Assembly Standing Committee on Labor and Employment

You should move in. You're like all the way over there by yourself.

Good afternoon, and welcome to our joint informational hearing on AI in the workplace. I would like to start by thanking privacy and consumer Protection chair Rebecca Bauer Cahan for attending today's hearing, as well as Committee Members, Assembly Member Irwin, and Assemblymember Dixon.

We will have other Members joining us throughout the hearing, and I will acknowledge them when they come in. Today's hearing focuses on AI's impact on workers. This year, we have considered dozens of bills related to AI, and it seems that everywhere we go, that's all we hear about is AI.

However, the one thing that we haven't really heard much about is AI's impact on workers. As chair of the labor and employment Committee, I want to ensure that workers voices are not left out of the conversation. And I want to make something very clear. I'm not opposed to progress or technology.

I'm hoping maybe one day they can have an AI version of myself. Being in three places at once is very difficult for many of us. But my concern is that with lack of regulation, I have watched the promise of AI turn Amazon warehouses into speedways that exhaust the workforce, and where stopwatches are the ones doing the firing.

We want to ensure that AI complements and supports our workforce rather than replace their Boss or even the job itself. Our witnesses today include workers, academic researchers, and industry experts who will share more about the landscape of AI. Today's hearing will be divided into three panels.

After each panel, we will have some time for Committee Members to ask questions, and at the end of their hearing, we will take public comment. Chair Bauer-Kahan, would you like to make some opening remarks?

Yes. Thank you, Madam Chair. I want to start by thanking you for convening this really important conversation today. And I want to thank, I want to reiterate my thanks to the Members of the privacy Committee that are here to join in the conversation.

Our Committee has heard, I believe, almost all of those artificial intelligence bills which you referenced, many of them that affect the workers that we hear from today. And also there's been an absence of bills that would help protect those workers as well.

So the bills we heard and the bills we didn't hear I think are important to talk about both today. I have long said that I believe that the potential of artificial intelligence, as chair Ortega said, is incredibly exciting.

You know, I mentioned often the ability for AI to detect cancer before the human eye can, that saves lives, that augments the work of those physicians and nurses that work so hard to save the lives of our cancer patients in California, the benefits.

I've heard from people in the educational space about using these technologies to augment their ability to teach students to read, that learn differently, that don't replace anybody, but help augment the workforce to make sure that everyone is getting met the way they can in ways that I think are incredibly exciting.

But that piece that I just talked about, I think is crucial to the conversations we're having today, and I think was at the heart of what Chair Ortega was talking about, which is augmentation rather than replacement.

What does it mean to use these technologies to better serve Californians but not disrupt our workforce in such a way that hurts Californians?

We're living in a moment in history where income inequality is at an all time high, where people are struggling to live in our communities because they don't have good paying jobs, benefits that help with their healthcare and keep them safe and healthy.

And the work that the labor and employment Committee does and that this Legislature does to ensure that California raises all of Californians wages, working conditions, and the ability to retire with dignity, benefits, et cetera, is critical to the work we do. And in any industrial revolution, we know it's something we need to pay attention to.

And as leaders of this state, this is something that we need to be talking about and working towards to ensure that as this revolution happens, workers are at the table. As Turo Ortega said, they are part of the conversation.

They are helping us learn how to implement this in a way that helps Californians and doesn't hurt those that will be affected by this work. Earlier this year, we held a hearing in partnership with the Arts and Entertainment Committee, focused in large part on the impact it has already had on their workforce.

And I know we have someone here from the string and Actors Guild to testify today, but it was incredibly powerful to hear the artists in Hollywood talk about the way their work has already been replaced, impacting their livelihoods, their jobs, their industries, and also the magical art that California believes and enjoys so deeply.

And so it's critically important that we continue to keep our eye on the ball as we protect California's workforce, as we protect California's economy, as we see this innovation happen, that those workers are part of that conversation. And so I really appreciate this lineup today and who we're hearing from.

I think it's also critically important that we see a partnership. You know, I do think that, that it's important that our workers and our employers are having this conversation together to move things forward so that workers aren't being surveilled in ways that harm them, that they have the dignity they deserve in their work.

And most of our workers, and I'm sure we'll hear this today, they love their work. They want to continue to do their work and they want to do it well, and they're proud of what they do. And we want to continue to provide an environment for every worker to do that.

And AI can be, hopefully, if we do this right, be a way to help that and not harm it, but to do it right, we need to learn and be on top of it.

So I appreciate chair Ortega and all of our staff from both the privacy community and the labor and employment Committee for their hard work convening this hearing. Thank you.

Thank you. Unless any other Members would like to make comments, I would like to call up our first panel. So our panel number one will be Doctor Annette Bernhardt, Director of the technology and work program at UC Berkeley Labor center, and Sarah Flocks, legislative and strategic campaigns Director for the California Labor Federation.

Zero, yeah, Doctor, she's on the. Doctor Bernhardt is on joining us remotely. Good afternoon, everybody.

Can folks hear me okay?

Can we hear okay? Okay. Yes, we can hear you.

Okay. Wonderful. Good afternoon, everyone. My name is Anetta Bernhardt, and I direct the technology and work program at the UC Berkeley Labor Center.

Doctor Bernard. Sorry. Let me, I'm going to pause you real quickly. I want to see if we can get the volume up in the room if there's a way of doing that.

Testing, testing, testing, testing, testing, testing, testing, testing.

Yes, we got it. Okay, now we are good to go. Yes.

Okay. Wonderful. Again, my name is Aneta Bernhardt. I direct the technology and work program at the UC Berkeley Labor Center. I really want to thank chairs Ortega and Bauer Cahan for the opportunities to, to speak with all of you today, and especially for your leadership in California's efforts to be a national model on AI policy.

For the past five years, my team and I have been conducting research and policy analysis of AI and other data driven technologies with the goal of ensuring that working families are able to thrive in the 21st century economy. And I'd like to share with you today several observations from our work.

So to start, it is absolutely clear that employers across the country are increasingly using data in algorithms in ways that tend to have profound consequences for wages, working conditions, race and gender equity, and worker power. And here are just a few examples. Hiring software now can rank job applicants based on their tone of voice.

During video interviews, algorithms are being used to predict whether workers will quit or become pregnant. Or try to organize a union call center technologies are analyzing customer calls in real time and then nudging workers to adjust their behavior. The retail industry is using just in time scheduling software, often wreaking havoc on workers lives with impossible schedules.

And, as has already been mentioned, Amazon's warehouse workers endure constant productivity tracking, pushing the pace of work to an alarming rate and putting their health at risk. I want to emphasize that these workplace technologies vary significantly. They range from simple payroll systems to task direction software to complex models predicting worker aptitudes and behavior.

And now, of course, generative AI. In my mind, the appropriate focus for regulation is this broad range of data driven technologies, not just one particular type. And in all cases, the core problem is that employers are introducing these often untested technologies with almost no regulation and oversight.

Currently in the US, workers largely do not have the right to know what data is being gathered on them or whether it's being sold. They don't have the right to review or correct their data.

Employers are not required to notify workers about electronic monitoring when they use algorithms to make employment decisions, and workers don't have the right to challenge those decisions. Most important, there are virtually no meaningful guardrail, meaningful guardrails on which technologies employers can use and how they use them in their workplaces.

As a result, there is broad consensus among legal scholars that existing employment and labor laws are inadequate to the task of protecting workers in the data driven workplace. Last year, workers in California did gain some basic rights under the CCPA, but those really are no substitute for the types of broad protections that workers have.

For example, in the 27 countries of the European Union. The critical point is that this regulatory vacuum in the US opens the door to a wide range of potential harms to workers.

I will say we are only beginning to understand how digital technologies impact workers, but there is already evidence of a range of harms, including work intensification and speed up, deskilling and automation, hazardous working conditions, discrimination, growth and contingent work lower wages loss of autonomy and privacy being blamed and penalized when the tech makes errors, and suppression of the right to organize.

And of particular concern is that workers of color, women, and immigrants both face direct discrimination via biases in the technology itself, and are also more likely to work in occupations at the front lines of technological experimentation. But I would say these harms are not at all inevitable.

I believe we currently stand at the juncture of two very different paths going forward. On the one hand, there is a dystopian path where employers use technology to cut labor costs by exploiting or displacing workers.

And there is also a high road path where employers use technology to augment and support workers, enabling them to gain new skills while also raising productivity. The policy choices we make today will determine which path we take.

And specifically, I would argue, we need a new set of 21st century labor standards that establish worker rights and employer responsibilities for the data driven workplace, much like the foundational standards that we laid down in the last century around wages and working conditions. These standards should be established both in public policy and in collective bargaining agreements.

You will hear more detail from the panelists that follow, but I'd like to lift up just some of the principles that unions and other worker advocates have developed over the past several years regarding AI and other data driven tech transparency.

Workers should absolutely have the right to know about all data driven technologies being used in the workplace, including AI guardrails.

We absolutely need robust standards to ensure responsible use when employers monitor workers, when they make decisions based on algorithms, or when they deploy technologies that automate or augment workers tasks, and in addition, use of unproven or questionable technologies, what are often called snake oil tech, should be prohibited.

Humans in command workers should have the right to override the systems they work with in order to prevent harms to themselves or the public. Right to organize workers should have the right to organize and bargain around new technologies.

Education workers should have a fundamental right to education and training so that they can adapt to changes in their jobs. Accountability AI and other digital tech should be tested for a broad set of harms before its use, including discrimination, deskilling, health and safety harms, and job loss, and finally, enforcement.

Government agencies should play a central role in holding employers responsible for any harms caused by digital tech, and workers should be able to sue for violations. So that is just to give you some flavor of the types of concepts that worker advocates and unions have developed.

A final point before I close, and that is that labor standards are absolutely critical, but by themselves are not enough. And ultimately, and you will hear this from others, workers need to be and should fully participate in decisions over which technologies are developed, how they are used in the workplace, and how the resulting productivity gains are shared.

This participation does not and should not be, need to be anti innovation because workers have a wealth of knowledge and experience to bring to the table. I truly believe, I think, like you all, that dehumanization and automation are not the only path.

With workers at the table, AI and other digital technologies can be put in the service of creating a vibrant and productive economy that is built on living wage jobs, safe workplaces, and race and gender equity. So thank you again for this opportunity, and I am happy to take any questions.

We will definitely have an opportunity for questions. I want to make sure that our next panelist goes, and then we'll open it up for questions of the Members.

So, Sarah Flocks, thank you, madam chairs and Members of the Committee. Sarah Flocks, I'm from the California Labor Federation, and I just wanted to thank you for having this very, very important and timely hearing. A in the fall, the Labor Federation convened a number of our unions on the topic of AI and technology.

And we had a day long session with Members from actors, writers, janitors, really, every person who is a union Member in California, and put together our principles, which are labor's principles, on artificial intelligence, automation, and technology in the workplace, and that I've made that available to the Members of the Committee here.

And Annetta and I have clearly been working together for a long time because she said, we have two paths to go on. And I have, in my notes, we are at a crossroads, and we really are. She said that there are two choices right now with technology.

We can go down the dystopian path, or we can go down the path where the gains of technology can be broadly and equitably shared. So I am actually going to go down the dystopian path because we have a history of that in this country and in this world.

And I think there are some lessons that we can take from that, of what to avoid. And so I'm going to go way back, like way back to 1811, when textile employers introduced the mechanized loom and knitting machines into the textile industry in England. And it gave rise to the Luddite movement.

And I'm going to give a shout out right here to Brian Merchant, who is the author of Blood and the machine, for going into this history. We actually had him speak to our Members and learned a lot from this. And so Luddites get a bad rap as being anti technology.

I've been called a Luddite because I wouldn't give up my BlackBerry. But really, they were not anti technology at all. What they were against was employers using machines to undermine basic labor standards. These were skilled craft people who wanted to see apprentices because they had an apprenticeship program, be the ones who were trained to use these machines.

They wanted to have decent pay. They wanted control over their hours. They wanted their pensions. They wanted to be able to have a family supporting wage and control over their work. Employers were not using machines this way. They put the looms into factories that really gave birth to the modern sweatshop.

They gave birth to dangerous, dirty factories where child labor was used, where workers were injured because there weren't safety precautions or training. And so that is the reason Luddite smashed machines, is because the employers would not listen to their demands, and they wouldn't negotiate with them.

And going a step further, employers decided to make it a capital offense to smash the machines. And so with that, they destroyed the Luddite movement. And we have lived with those consequences for decades and decades.

And it has taken us blood, sweat, and tears to get a ban on child labor, to eliminate sweatshop labor, to get a minimum wage and a weekend.

All of that was fought for because there was not regulation when a lot of these tools were introduced into the workplace back in 1811 or back in whatever period where technology was brought into the workplace. So technology is nothing new. And unregulated technology is nothing new.

However, AI enabled technology poses a whole host of new challenges that I think it's very important for policymakers and union Members to understand, so that we're able to sit at the table and bargain for the right things, and that the important guardrails are put in place, because the guardrails and standards that Annetta talked about, they're basically what the Luddites demanded in 1811.

And so the foundation is already there. We just have to adjust for some of the changes. So one of the major changes is the rapid advancement and the corporate investment in AI technologies. Who knew about AI five years ago? Like, it is advancing at a rapid pace, almost too fast for us to understand.

And it's not a loom. It's not something we can touch. It's an algorithm. It's a black box where we don't necessarily know what goes in, and we can't, right now, take it apart. And so that is a real challenge.

The other thing is that we say or we think that technology is neutral, but it's not necessarily, because look at who is developing and producing it. Corporate giants like McDonald's, Walmart, Amazon are investing billions and billions of dollars into this technology so that they can do workplace surveillance, so that they can maximize productivity and cut labor costs.

That is how the technology is being developed for that purpose. We also have VC venture capital, who is investing, I think it's $290 billion into AI technology over the last five years to develop things that they can make money off of, not necessarily that makes work safer or more productive or beneficial for workers.

So we need to not just think about how technology is deployed, but how it's developed and what are we investing in the development of technology that is for workers? And that is a piece there that's very important.

It's something that the national AFL has looked at as well in how do we understand what technology can we develop that helps nurses, that helps doctors, that helps make hotel housekeepers jobs less backbreaking? The second piece is that AI and the data that enables AI has supercharged existing exploitation and worker surveillance.

And so while cameras and bosses trying to spy on workers has always existed, the ability to collect data has just taken off.

When bosses can track your eye movements, when they can use software to know when you're typing, when you're clicking on your mouse, when they can use all of your data to do predictive behavioral analysis that Annetta pointed to, where they, when they're hiring or looking at a workforce, there is technology where they can say, is this person, are they likely to file a wage complaint?

Are they likely to unionize? In fact, there is technology that exists, and I have a list of it, perceptics, which is used by 30% of Fortune 100 companies. They have something called a union vulnerability index that can rate how vulnerable workers are from just data they collected to joining a union.

Workplace by Facebook allows employers to suppress and monitor the term unionization in chats between workers. Humanize. I really can't believe it's called humanize. It monitors workplace conversations between workers. And terramind tracks the written communications and triggers notices to bosses when keywords are used, such as unionization. And there's a whole bunch more. All of these are spelled incorrectly.

It's humanized with a y, just like tech likes to do. There's Oracle retail Xbri that monitors cashiers daily and does a high risk assessment of who's most likely to steal. All of this is in a black box. We don't know what's going into this box to figure out which worker is high risk.

So the monitoring and then algorithmic management where there's a haptic system that Amazon uses. So if you're in a warehouse, they buzz your arm to move. So you're the most efficient you can be. It's like scientific management, as if you were like a robocop. This eliminates any micro breaks, any moments that workers would be able to rest.

The rate of injury in Amazon warehouses is extraordinary because they are basically treating workers like robots and they have the technology to supercharge that. So I have so many more examples, but I know, I think I'm at time so I will also say pervasiveness. This technology is everywhere.

I have never had an issue where literally every worker is impacted. Ones that I'm surprised. It is everywhere. It is touching everyone in all different ways.

And it's not only an impact on workers, but it's on the public, the people they serve, patients, every one of us who walks into a grocery store and has all of our data collected and our retina scan to see do we look at Ayes cream for too long? That means we get ads about it.

And so the pervasiveness in our lives, and yet how invisible it is and how little we understand poses new and unique challenges. So we have a choice, like Annetta said. And I would urge the policymakers here to be smart, strategic, and very thorough in the regulation of technology as it's developed and deployed.

And you have such a great panel. You're going to hear from so many union Members who have experience and who've bargained this. There's a lot of great examples in collective bargaining. So again, just thank you so much for this hearing and this opportunity.

Thank you. I will now hand it over to the Members. I have questions, but if other Members have questions. Okay, actually, I'll wait because one of my questions was going to be around examples of things that are happening. The next panel. All right, thank you. Thank you, Doctor Bernard.

We're going to move on to our panel number two. So we have NYU Professor Shellman, who I believe is going to also be remotely Rosalba Arcanega, janitor with SEIU USW, as well as her translator, UCLA Professor Chris Tilley, who's also going to be remotely.

So are we going to have both of them on the screen or they're going to take turns. Okay. Zach Gaynor, campaign researcher for UFCW local Five. Chris Nielsen, Director of Research for the California Nurses Association. So this is a pretty big panel. We're going to limit it to six minutes for each panelist.

And then again, once everyone is done, I'll turn it over to our Members for questions. We will start with Professor Schellmann.

Yes. Yes. Hi, everyone. Thank you so much for having me. Let me know if you can hear me well or see my own thumbs. Yes, I can hear you. It's good. Perfect. Wonderful. Great. So, hello from New York. So my name is Hilke Schellmann. I'm an Emmy award winning investigative reporter and assistant Professor of journalism at NYU.

And for most of my professional life, I've worked at the Wall Street Journal. I'm speaking to you today because over the past seven years I've been reporting on artificial intelligence and hiring. I wrote countless articles and recently published a book on the topic.

I've talked to over 200 job seekers, experts, HR managers, recruiters, hiring managers, psychologists, lawyers, basically everyone and anyone who has witnessed how AI has fundamentally changed the world of hiring. And I wanted to thank you all for your interest in this I think, urgent topic, and I wanted to share what I found out.

So, first of all, the use of AI in hiring is widespread but underreported. So approximately 97% of Fortune 500 companies use algorithms in their hiring funnel, from checking and rejecting resumes to video interviews to AI powered background checks.

Almost all applicant tracking systems that many mid sized and large companies use to track job seekers like where they are in the hiring phase, have AI now built in. Job platforms like LinkedIn and indeed, and others also have AI built in. And no one is checking if the way they're using it is leading to discrimination.

In 2021, the CEO of the job platform, Ziprecruiter, Ian Siegel, told me that between 75% to 100% of resumes are currently checked by AI, noting that that is three years ago, it's probably more now. Companies use AI tools because many get thousands or even millions of applications per year.

Since the dawn of job platforms, again, like indeed in LinkedIn, made it much, much easier for job applicants to apply to open positions with a mouse click. So on the side of the companies, they say they're inundated by resumes and do not have enough capacity to check all of them.

Google, for example, gets around 3 million applications a year. IBM around 5 million. Goldman Sachs got over 220,000 applications for their summer internship program alone. And also, AI is used in one way video interviews, in which candidates record themselves answering questions with no humans on the other side.

And one of the leaders in one way video company interviewing does about 100,000 or so video interviews every day. Most job seekers I've spoken to didn't know that AI was being used on them.

And even if they knew, they said if they really wanted the job, so most of them would have not opted out of the AI screen. And here's some of the problems I found with the different technologies that are being used. So AI is pitched by the companies as the great equalizer.

Hiring one objective algorithm measures them all without bias. So I was hoping to find evidence of this. But what I found is a little less hopeful. For example, I spoke to three employment attorneys and one psychologist who got access to look into different black boxes of resume screeners.

They told me that they found problems in many of the tools they were asked to check. Some tools used first names like Thomas as indicators of success, which means that job seekers who had the word Thomas on their resumes got more points, which obviously has nothing to do with someone's qualifications for a job.

Even more problematic keywords were Syria and Canada, which may lead to discrimination based on national origin. Another tool gave more points to candidates with the word baseball on their resumes, and fewer points to candidates with the word softball on their resume, which indicates gender discrimination.

In the US, another tool used the words Africa and African American, which indicates discrimination based on race. These type of AI tools often go unchecked because HR departments buy them to save money and do not want to hire more people to oversee these tools.

Also, some company executives have told me that they're not checking these systems so they will have plausible deniability if there's litigation. AI is also being used in video games for hiring, and one employment lawyer told me that a tool he tested kept discriminating against female applicants.

The startup that built the video game suite is still selling the same tool and has even grown. I played some of these games as well, and in one game, I had to hit the spacebar as fast as possible. While I was doing that, I was wondering, what does I have to do with a job?

But also, what would happen if an applicant has a motor disability? Would they be rejected by the AI because they couldn't hit the spacebar as fast as I could? Is that fair? I reported in 2018 for the Wall Street Journal that some of the largest video interview companies used AI that scanned the facial expressions of job seekers.

It turns out there's no science of what facial expressions we need for a given job. And there's no science for facial expressions in job interviews that predict success in a job. And we see this again, that we use technology not based on science. Luckily, public pressure had gotten to the company and they stopped using the tools.

But other companies still use it. AI is also used in background checks for employment. Some tools analyze social media feeds of candidates, and this broad kind of analysis could find protected evidence of union organizing amongst workers.

As I said, I've also tested some of these tools myself, one that was supposed to find out how well I speak English. I actually spoke to it in German, which is my native language, and I got a six out of nine English competence score. That was interesting.

So I talked to the developers of this, and a lot of them who built these tools don't know themselves what exactly their tools are doing. And I think that strikes me as very problematic. People who built the technology don't know what the technology does, and folks in the industry actually call all this the wild west.

And just to sum it all up, the EOC and other enforcement agencies have a hard time checking these tools because they work based on complaints from job seekers.

But since most job seekers don't know what is being done to them, they don't even know AI is being used on them, let alone that they may have been harmed, there are very few complaints.

So I believe urgent action is necessary since we know that there are problems, especially for people from already marginalized groups and people with disabilities are not being considered when one size fits all tools are being built. I believe fairness and access to the labor market based on merit is at stake here.

We need transparency, explainability, and more research. Thank you for your time and I'm of course happy to answer any questions or give anyone more information.

Okay?

Rosalba yes. Presidente. I assembled gracias, rosalva Tango. Just put it closer to I'm sorry. Okay. Okay. Thank you, Presidente.

Assembliesta gracias por laportunidade testificarante este importante comite esun honor parami Min hombre ezo salva tengo onse anus en LA industria De Olympiesa Gran Parte demi carrera esta et industria etra bajado en LA ciudad Sanibel, California haciendo LA limpiesa enuna De Las empressas technological soyori hinaria De Mexico mi Familia emigroa estepais bus cando LA reunificacion Familiar Olympiesae Siempre Avenue amasa Las companias que Nosotros comoyanitorios por maz De sanos parapoder ponchar entrada Salida tenia que bahar una applicacion enmi telefono Celular lampresa Nos Dava Quattro Dolares almes Como rambolso aviadia scale o Companeros Que Noles Nolesa Justabel Dinero Parapagar sms e nostocava sufrir tratando De esplicar Al Supervisor porque no podiumos ponchar Aurora Veo Ken El Valle De silicon irobots, portolo lados and restaurantes Iotras industrias sece LA intelligence artificialidad Esquempo Para que mas y mas impacte LA industria De La Limpiesa Sequela janitorio robotos Di senados por Las mismas impresas milionarias que limpiamos es incre ible que Las grandes empresas Imbierta La industria De Lampiesa depende De La esplotacion De Las trabajadoras De La olympiesa ensu Mayoria Somo mujeres immigrantes.

Las que Limpiamos Comoyanitorio y mujer immigrante conanos Esperiencia e nesta industria puerto atesti guarantee LA sobrecarga De trabajo. Ilo ilagotador De Nuestra Labor munchos De Mis Companeros De Trabajo e nesta industria dependent De Medicamentos Para El Dolor Para Aliviarde LA tenciones Del Travajo.

Yal Unos Incluso Ancido somatidos a sirujia por leciones que persistent en apes De La intervention medica De Esto De sardestos De safillos segimos trabahando in can supplemente paramantener LA limpiesa De La De Los edificios mass prestigiosos.

Amenudo Conun Gran sacrificial personnel Que Bueno que seuse pero importante que LA intelligence artificial noes Utilizada para desplazar altra bajador De La olimpiesa LA technologia De Beser El Pueblo Debe acabar con les plotacion y LA carga laboral.

La intelligence artificial editores De Programmar robots De dar opportunidad De crecimiento en La industria e De gurantizar Travajos bien remunerados dondes tratados.com respecto idignidad La intelligence artificial no puede acer el Travajo De Unumano. No ac muncho tiempo en medio De La pandemia mass. Or De Humana Los Ricos De Sienna. Gracias.

Thank you. I'll be translating for rosalba. I want to clarify, I cannot translate for her, but I can read English, so I appreciate chair Ortega also offering her staff for if there's any questions. Thank you, madam chairs and Members, for the opportunity to testify before this important Committee. It is an honor for me. My name is Rosalba.

I have 11 years of experience in the cleaning industry. For much of my career in this industry, I have worked in the City of Sunnyvale cleaning at one of the largest technology companies in the world. I am originally from Mexico, my family emigrated to this country seeking family reunification.

I feel very proud to be a migrant woman and to be a janitor with a union. My union is SEIU USW. The use of artificial intelligence in the cleaning industry has always benefited companies more than us janitors.

For more than 10 years, to be able to punch in and out, I had to download an application on my personal cell phone. The company gave us $4 a month as reimbursement.

There were days when the cell phone didn't work or colleagues didn't have the money to pay for it that month, and we had to suffer trying to explain to the supervisor why we couldn't clock in. Now I see that in Silicon Valley, there are robots everywhere, in restaurants and other industries.

I know that artificial intelligence is already a reality, and it is a matter of time before it impacts the cleaning industry more and more. I know there are already robot janitors designed by the same $1.0 million companies that we clean. It is incredible that large companies invest millions of dollars in developing artificial intelligence.

And every time we ask for better salaries, there are always excuses for not paying us decent wages. The cleaning industry depends on the exploitation of cleaning workers. It is mostly immigrant women like me who clean.

As a janitor and immigrant woman with years of experience in this industry, I can attest to the overworked and exhausting nature of our work. Many of my coworkers in this industry rely on pain medications to relieve the stresses of work, and some have even undergone surgery for injuries that persist despite medical intervention.

Despite these challenges, we continue to work tirelessly to maintain the cleanliness of the most prestigious buildings, often at great personal sacrifice. I am not against artificial intelligence. It is good that it is used. But artificial intelligence mustn't be used to displace workers. Technology must serve the people. It must end exploitation and the workload.

Artificial intelligence is already here. This is the opportunity for technology to give janitors the opportunity to program robots, to provide opportunities for growth in the industry, and to guarantee well paid jobs where we are treated with respect and dignity. Artificial intelligence cannot do the job of a human.

Not long ago, amid the most horrible pandemic in the history of our human species, the rich said that we janders were essential. Now the rich want to replace us with machines, and that is not right. Thank you very much.

Gracias. Thank you. Next we have Professor UCLA Professor Chris Tilley, who is also remote.

Hi, I'm Chris Tillis. I'm a professor of urban planning and sociology at UCLA. Glad to have the chance to address all of you. I have a PhD in economics, and for 40 years I've been doing research on bad jobs and how to make them better.

I've studied retail jobs in particular, on and off since the late 1980s, and starting in 2018, I've been conducting research on how technological change is affecting store based retail jobs. This recent research includes about 901 hour interviews, 110 Shorter, 15 minutes mini interviews with retail professionals, including executives, technology providers, and consultants.

In my testimony today, however, I'll mainly be drawing on 631 hour interviews with store based employees and managers conducted over the last two years. These were mainly workers in chain stores, in grocery and General merchandise, including a lot of big box workers all over the country.

But a good chunk of these interviews were with California workers, and a lot of the retail technologies that are talked about are getting piloted in California. I'm also going to refer briefly to numbers from the Bureau of Labor Statistics. When it comes to understanding how technology affects jobs, the devil is really in the details.

So I'm going to focus on three particular jobs. First, cashiers. Between 2019 and 2022, which is the most recent year we've got data for now, the number of cashiers working in grocery stores fell by 14%, whereas the overall number of workers in grocery stores only fell by 3%.

In General merchandise, which includes retailers like Walmart and Target, the number of cashiers went down by 23% at a time when total employment in those stores was growing, and by 3%. It's very clear this is due to the spread of self checkout. The interviewees talked repeatedly about how self checkout kiosks were replacing cashiers.

One Walmart manager commented that the quote plus of self checkout is less payroll. We heard about managers watching up to six kiosks, and most of them dislike that job.

There's also new technology based surveillance of cashiers and customers, like Walmart's missed scan system that uses overhead cameras and warns cashiers if the system detects that they missed scanning an item. And also, of course, alerts management. The next job is online order fillers. A huge number of online orders get filled in stores for pickup or delivery.

Target, for example, fills most of its online orders in its stores. Here, the level of surveillance is overwhelming. Workers pick orders using a phone or other handheld device or smart glasses, and the system basically keeps track of every move they make, with managers tracking how quickly and accurately the work gets done.

Walmart tracks five main picking statistics, and at least some Walmarts post all the pickers scores publicly. Finally, I want to talk about stalkers. There are increasing number of cameras all over the store, mounted on shelves, on robots that roll around the store, hanging from the ceiling, even on drones.

Amazon created the just walk out technology, where you put the merchandise in your basket and just walk out of the store without checking out, which may sound like magic, but they've got hundreds of cameras hanging from the ceiling in an Amazon fresh or Amazon Go store that uses this technology.

Retailers have also shifted from having people watching the camera feeds to what's called computer vision, meaning AI based software that can interpret the images and alert a manager or worker or both if something is not going according to plan. In most cases, this includes facial recognition software, which is illegal in Europe.

It's not clear how much retailers are using all this camera technology to keep an eye on workers. So far, managers are very tight lipped about it, but a good number of workers think they're being watched. And what's very clear is that the technology makes it possible to watch both shoppers and workers again.

That could be people doing the watching, or it could be computer vision software, with all the kinds of problems of software that have already been discussed. One other important point is that technology often does not work correctly.

I heard about just about every possible screw up with some of these new technologies, and in some cases, these glitches are happening weekly or even more often. I would single out three areas where public policy could really make a difference. First is requiring a technology impact assessment for major technology changes.

Retailers are implementing new technology at a pace that is outrunning our understanding of its implications. An impact assessment will help us monitor for potential downsides. This imperative is particularly strong for service industries like retail, where technological changes impact customers as well as workers.

Second is to lay down some worker and consumer transparency and privacy rates related to new technology. We can decline advertising cookies, but we can't decline facial recognition software in this store. Again, the European Union has very strong restrictions on facial recognition software.

And a third, more retail specific issue is to place some limits on the use of self checkout technology, which is a job killer and its current form is making both workers and consumers miserable.

Retailers have their doubts about self checkout, but for as long as I've been studying the retail sector, it's been dominated by a Low cost, Low service, high labor churning model where if retailers don't follow the crowd, they'll get undercut by competitors.

It's in the public interest to close off competition that degrades work and the shopping experience in pursuit of extreme cost cutting. So I'll leave it there. Thanks.

That was impressive. Yeah, that was impressive timing. Thank you. Next we have Chris Nielsen, Director of Research for the California Nurses Association.

Yeah, thank you.

Zero, did I skip? Zach? zero, I apologize, Zach. You were hiding back there. If it's okay, I'll go to Zack Gaynor, campaign researcher for UFCW local five.

Good afternoon, chair and Committee Members. My name is Zach Gaynor. As a chair, sorry. And as the chair said, I'm a researcher with United Food and Commercial workers local five. We represent 30,000 workers in Northern California from Salinas to Eureka and the entire Bay Area.

UFCW represents private sector workers in grocery, as we mentioned, drug, retail, retail, food processing, agriculture and cannabis. Technology and automation of workers jobs is not something that's new. We've talked about it, but as a union, we strongly believe that any new consequential technology deployed in workplaces needs more worker centric and involve active input from workers.

Utilizing the technology be used to make workers jobs safer and not replace the jobs of workers.

Unfortunately, oftentimes in the case is used to eliminate workers jobs, which has led to significant challenges in stores for workers, customers and employers, especially with the implementation of so so technology under the definition of sorry, Restrepo is basically, and I'll read this sorry technology that lowers the demand for labor without any strong gains in consumer welfare or productivity, threatening long term employment and wages.

And so I'll talk about self checkout and how that fits that definition. But employers are quick to assert that AI and technology is not replacing workers jobs, but freeing workers up to perform other tasks, like providing better customer service. While in theory that sounds great, it's not the case.

In reality, there used to be eliminate jobs, forcing workers to work with skeleton crews and unsafe staffing levels. It's a net negative when it comes to employment in the grocery industry.

As the Professor said in stark numbers, the UFCW represents workplaces that are largely open to the public and staples to the community to get essential goods and services. You can't go. You need to go to a grocery store to get food.

When technology is deployed in our workplaces, it's not only deployed on the workers, but the customers as well, but they get limited notice or information about that technology before it's deployed. So self checkout, basically, even though self checkouts have been. Even though self checkouts are billed as automation, they're pseudo automation at best.

It's basically a transfer of labor from the workers who are paid to work there to the customer. And so you're not actually automating anything, you're just transferring the cost of labor to the customers.

And instead you're changing the job fundamentally from one of a cashier, where I went to a lot of different cashiers, and one of the things they relish is talking to the customers, being able to have that one on one interaction while they're scanning their food and talking to them about their day, their life.

I had people go to specific cashiers because they'd gone to them for years. And instead of having those one on one interactions and being able to develop relationships, they have to deal with up to six to 12 machines at a time.

Which means when you're going to talk to a customer, you're going to someone in distress, not somebody who's coming to talk to you about what's going on in their lives. But hey, this says that something's wrong with the machine.

And so your job fundamentally changes from one of talking to customers to almost dealing with the machine, which is a different, which is a machine that's also replacing your labor.

So there's double indignity in that the machines are notoriously glitchy and not accessible to all customers, including those with disabilities, non English speakers, customers requiring additional assistance, and the unbanked, which will cause further problems with some of the new technology that's being proposed.

Furthermore, new self checkout technology requires fingerprint scanning, retinol scanning, new smart card technology requires customers to have credit cards and smartphones, which will be a significant problem for those who are unbanked or use EBT. EBT has very strict requirements. You can't just refund it.

So if these carts make mistake and you put it in and it scans it, there's very little workaround. And especially if you cut out all the labor around it, who's gonna help? Who's gonna deal with that problem? It creates a range of problems for our overworked and understaffed workforce. Loan.

Self checkout works are expected to monitor anywhere from four to 12 machines while performing other tasks like processing money orders, selling lottery tickets, opening locked cabinets, and handling customer service phone lines. It leaves very little time for our clerks to actually assist customers when self checkout machines malfunction, handle actual customer complaints, or respond to irate customers.

The lack of staff at the front of the store has led to a massive increase in theft. We don't know the numbers, people are being pretty tight lipped about it. But over 20 million Americans have admitted to stolen something at self checkout and nearly 9 million Americans sharing, they would do it again.

Retail theft at self checkout has led to retailers losing over $10 billion in gross revenue. In addition, 67% of consumers reported dysfunctional self service kiosks. 418% of consumers avoid self service kiosks because they experienced a slower checkout process, and 20% of consumers avoid self checkout because they preferred interactions with employees.

This is just one example, but the examples that have been mentioned that might be implemented would have severe consequences if implemented without worker input. So surveillance cameras and facial recognition technology. We're starting to see the deployment of surveillance cameras in grocery store with facial recognition technology that impacts both the consumers and the employees.

But as we've seen with many different stories of facial recognition technology not being able to properly identify people, especially people of color, that runs a major risk at a grocery store where if you're wrongfully profiled and detained, that could lead to you not wanting to go. And where else are you going to go?

I don't necessarily have a ton of time to go into other technology, but there is being electronic shelf tags, which would lead to surge pricing, basically the addition of surge pricing in grocery stores where they could change the, as they say, the price of Ayes cream on a summer day so they could charge based on demand.

So basically, we support technologies that have, we want use case studies, we want worker input, and we want to support Senate Bill 1446 by Senator Smallwood-Cuevas that would address the fallout of self checkout as well as some other technologies as well.

Thank you, Mister Nielsen, you may go now.

Sure. Thank you so much. Thank you, chairs. Thank you, Committee Members, for the opportunity. My name is Chris Nielsen.

I'm the Director of education for the California Nurses Association and here on behalf of over 100,000 CNA nurses across the state to share some of our experiences and our concerns about AI and healthcare, which poses some unique concerns and risks. Nurses have always embraced new technologies that enhance their skills and then improve patient care.

But we're concerned that in the current environment, the rush to deploy unproven and still mostly unregulated AI tools in healthcare settings, we really risk undermining the humanity at the heart of nursing care and putting patients in our communities at risk.

Back in March of this year, our union surveyed more than 2300 rns and found that 60% did nothing. Trust their employers to implement AI with patient safety as the first priority.

Long before this current generative AI boom, we've seen employers deploy more conventional AI and algorithmic tools in ways that undermine the human expertise and the clinical judgment that's central to the work of healing.

Algorithmic technologies drive all kinds of tools in hospitals and healthcare settings, from the clinical decision support algorithms that are embedded into the electronic health records that nurses use at the bedside. They show up in remote patient monitoring systems and predictive staffing tools, and many of these tools are prone to serious inaccuracies.

And yet, 40% of the nurses that CNA surveyed said that they do not have the ability to override these AI recommendations, even when they know that they are incorrect. So let me give you just a couple of examples of some of the issues that we are seeing.

Kaiser and many other large healthcare employers use the epic electronic health record system, which often includes a tool to determine a patient's acuity or how critically ill they are, and then generate staffing recommendations based on these scores.

But these black box algorithms work off a limited set of data, and they often missed a the nuances and subtleties that a registered nurse would be able to assess for the accuracy of these tools also depends on the nurse's ability to chart in real time. But in chronically systematically understaffed hospitals, this is often impossible.

So it's no surprise that two thirds of the nurses that we surveyed said that these automated acuity scores do not match their own assessments of a patient's acuity. And these inaccurate assessments can lead to harmful delays in treatment, potentially even deadly adverse events.

Algorithmic bias is also a major problem in our workplaces, and nurses are concerned that it could be exacerbated by this rush to deploy unproven generative AI systems, the newer kind of generation of AI that we're seeing.

A recent Stanford study of four generative AI tools that are being piloted in hospitals across the country found that all four of them perpetuated debunked race based medical practices based on the misconceptions about the supposed biological differences between patients of color and white patients.

And I think we're all probably familiar now with generative AI's propensity to hallucinate, in the industry parlance, or just make things up, which in healthcare, as you can imagine, would pose unique risks.

A recent study of a generative AI tool that's being piloted at a hospital in Boston found that it made safety errors 42% of the time when responding to simulated questions from cancer patients.

And here in California, several employers have begun to pilot similar tools for a range of use cases, from listening in and summarizing patient visits to automating clinical documentation and responding to patient communications.

So again, just using Kaiser as an example, they've begun using a tool called desktop medicine to read patient messages and then automatically route those messages to what it thinks is the right clinician to respond.

But that's the role of a triage nurse who uses their clinical judgment to assess a patient's needs, to ask additional questions if necessary, and then get them the appropriate care employers. One of their common claims is that generative AI in particular, will save nurses time on administrative tasks so that they can focus on direct patient care.

But in our experience, this is not the case, and some of the emerging studies bear this out. A study recently from UCSD found that a generative AI tool being used there did not save Clinicians any time when they were using it to draft replies to patient messages.

And in fact, they spent 22% more time responding to messages because of the additional time needed to review these drafts for errors.

So the point is just that the emerging science around generative AI in healthcare does not support claims that it can substitute for the clinical judgment or the effectiveness of human nurses, and attempts to do that pose serious threats to patients, to communities, and to our profession.

The good news is that through collective action, through collective bargaining, nurses have been able to push employers to improve some of these systems, make some of them safer, more transparent, or in cases where it's not possible to deploy these systems safely to prevent their implementation altogether.

But we also need public policies that empower healthcare workers to be in the driver's seat when it comes to AI. We need to strengthen workers rights across the state to bargain over these decisions, not just over how they should be deployed, but whether certain tools should be deployed in certain use cases at all.

We believe that legislation should be based on a certain set of principles. And I'll wrap up here momentarily. We've released some principles in our nurses and AI patients Bill of Rights, and we can share that with you all after the hearing. And I'll just conclude by highlighting two of those principles.

First, all patients deserve the expert, compassionate, hands on care that only human nurses can provide. We've seen unions in other industries, SAG AFTRA, the writers Guild, demonstrate that it's possible to draw a red line when it comes to replacing workers with AI.

And we need public policy in healthcare to allow nurses and workers to do the same thing. And in the case of other potentially acceptable uses of AI in healthcare, uses that don't replace or deskilled nurses, we need regulations that are based on the precautionary principle.

Right now, we have essentially a wild west in healthcare, where AI tools are being deployed without any evidence of their safety, their efficacy, equity.

So we need to be able to put the burden of proof on to tech developers, to deployers, to employers to prove that these technologies are safe, effective, and equitable before they're used in real world healthcare settings. And, you know, we do, as others have said, have the opportunity here in California to set the standard.

And we look forward to working with all of you on policies that can put our patients and our communities first. Thank you.

Thank you. I would like to recognize that we have Assembly Member Wendy Carillo, who has joined us, and Assembly Member Brian, do you have any questions for this panel?

I really appreciate hearing from everybody. We are in all these areas in the Wild west trying to figure out what direction we go to really benefit workers and patients.

And as Assemblymember Barakan said, we really, the idea is that you're using AI as a tool, but it is disturbing to hear with the nurses that there are AI tools that you can't override. I mean, that's not using AI as a tool to augment your work.

So do hospitals have, do some of the hospitals have policies that there always has to be a worker in the loop?

Yes. Well, I think it's important to maybe differentiate between a worker in the loop after these tools have been implemented. That's what I'm talking about.

That is communicating with the patient.

Yes, yes, yes, yes. I think that's important. It's also important to, I think, include workers earlier in the process. Right.

And we have negotiated contract language with a number of employers that requires notification, requires the right to bargain over the deployment, the design in many cases of these tools, and that importantly, safeguards the clinical judgment of nurses to be able to override. Now, that exists only in these facilities where we have those protections negotiated.

And so I think part of the task here is to expand and extend those types of protections to nurses and healthcare workers across the state.

And then I guess for all these tools, I think the really important, the other really important responsibility of the employer is to continue to upskill the workers so that there is always a job for the worker. And so are we seeing that in bargaining yet?

They haven't had the opportunity to bring this up, but if they ever propose this, they would be more than happy to carry this conversation on.

You're using the word deskill, and I think that that is, we cannot have that going forward. We should only be talking about the responsibility to upskill.

I think we do have upskilling as a part of the conversation, but it's also understanding how the technology would be rolled out, which is why one of the key things that people, I think here today are enhancing is that workers need to be in the conversation early so that instead of being deskilled, so they need to be upskilled, the technology can actually be used to supplement their work so they're not being deskilled.

And we're having to come up with new ways to upskill these workers when they can keep doing their job and the AI can support them rather than undermine them. So there's two ways of approaching it that we are trying to consider and definitely have in negotiations.

But I think a key component is having the workers engage in the technology early. So that's assisting them and benefiting more than deskilling them.

Right. Well, being able to be in the loop early, but also being able to override and use judgment. Thank you. And again, a great conversation. Really important.

We've had a lot of, basically, the use of the implementation of technology displaces workers to what. What they will then either negotiate at lower rates or will try and exclude from the bargaining unit. And so we've seen it mostly used to push workers wages down and to thin out the schedule.

In most of the retail industry, especially with grocery, you're dealing with food. There's perishable items. It's not necessarily. And machines also can't, because machines can't always grab. So I know it's more technical, but a machine can't grab broccoli without bruising it. So you create, again, you're just shifting. You're shifting the labor to another, another type of job.

And then they're declaring that jobs less valuable and then paying people less. So there are sometimes new jobs, because, again, this most of its displacement of labor.

But we need to figure out ways to make sure that workers don't keep losing against corporations that have the ability to just implement this technology without any input whatsoever from workers or anybody else, quite frankly.

Chair Bauer-Kahan and then Assembly Member Carillo. Thank you, Madam Chair.

I really appreciated what Miss Arenega. I got that right, what she said about the janitors being partners in training and using the technology. And I think it really goes to what we heard Miss Irwin say, which is, how do we make sure that our workers are at the table?

There was actually an article, I think, just this week about, and I know we don't have any of our education folks here at the table, but obviously, they're engaged deeply in this conversation as well, about the San Diego school district bringing in AI to grade papers and how inaccurate it was.

And when I read the article, the article actually didn't mention whether teachers were at the table when that AI was selected. But my gut tells me I know the answer.

And I think that that is a perfect example of where if you put the people who know this the best, which is our workers, at the table to make selections to help decide where it's appropriate and how to use it, we wouldn't have these problems. Right?

So I think it also is critically important that we're smart about this. And in so doing, I think we can only do that by bringing workers to the table to figure out how it best augments their work in ways that are effective and not detrimental to the end product here, the students.

But I also, you know, and I did want to bring up the consumer piece, which I think is really important because I think a lot of what you're advocating for is both important to our workers but also important to the consumers, some of which was mentioned briefly.

But I recently took my children to the pediatrician and was handed a release form. The release form was for an AI tool that the Doctor wanted to use to do their note taking.

The release form said that the companies that owned the AI, among those was Amazon, would have the right to use my voice and any information taken from that AI in any way and redistribute it without HIPAA protections.

I was so appalled that I not only didn't sign the form, but I went to my Doctor and I said, I can't believe you're even asking people to sign this. Now, I'm an AI policy expert, right? So I read that and I know what to do with it.

But the fact that we are even asking patients to sign release forms that give away what we, as the government have deemed the most critical private information in our lives, our health care information is a yemenite, appalling to me.

But that is what, in this instance, I'm assuming, Amazon's lawyers had handed to the healthcare company and then trickled down to my Doctor, who cared deeply. And when I brought it up with her, she didn't actually realize it was in the release and was concerned as well.

So I think the consumer's really important here and again had my Doctor, who was the worker in this instance, it can be nurses, it can be been brought into that conversation. And if we put this in the Doctor's office, your patient's healthcare information will now no longer be HIPAA protected.

I actually trust that the doctors would have made a different decision because I do think the workers often make decisions that are more protective of patients, consumers, students, whoever it is in the setting in which AI is being used.

And so I think it's such an important point that was brought up today, and I really wanted to elevate even further. But I did have a question about, you know, I think something that has come up today that is really important is this conversation on impact assessments.

And as many people in the room know, I've been working on a Bill dealing with much of what we heard about as it relates to bias and some of these automated decision tools that's surrounding this question of doing impact assessments and ensuring that bias is removed from those tools before they're deployed, also providing notice so people know when these tools are being used and can opt out, etcetera.

But there is really this interesting conversation around how do we make sure that as we put this regulation into place, it's effective?

We want to regulate, we want to protect workers, we want to protect consumers, but we also want to make sure that we do it right and then we don't just put regulation out there that isn't as effective as we need it to be for consumers. And I will say that, that our constituents want us to do this.

I've seen polling now over and over again that shows that somewhere around 90% of Californians want us to regulate artificial intelligence. They don't like the way they view it as going to affect their jobs, their lives, their privacy, name it, the environment.

We could talk about the energy usage of AI and the way that's going to affect climate. There's a lot of impacts that people want us to be focused on.

And so I think we at this table, the majority of us, I would say, want to regulate this in a way that really helps workers and consumers, which I think are often aligned, but we want to do it right.

So I was just wondering if you had any thoughts on sort of what is most important for us to pay attention as we move forward in regulating this. Because I think it's important that we keep those two things central. And to me, putting workers at the table is probably the most obvious.

So I guess you don't need to say that again because I've made that. I've reiterated that point. But anything else that you think we should really pay attention to. Thank you.

Could you repeat the last question?

Yeah. What as policymakers, beyond putting workers at the table? Because I think that point is well taken. Should we be most, should we be most aware of as we move policy forward.

I'll let the professor go. He can go first because he's been...

Yeah, it looks like the professor has his hand up on the screen, so go ahead and call on him.

There's actually two professors with their hands up, and I think the other one had her hand up first.

That's okay. I'll just quickly reply. So I feel like the consumerism is actually something that's really under, like, under thought in all of this. And I call this actually, for AI and hiring these tools, I call it forced consumerism because everyone who wants the job, are they going to say no if you get a link and says your application will be thrown out unless you do this AI based video interview? Even though oftentimes the AI part is in fine print and a lot of job seekers don't even know.

But when I talk to them, they're not going to say no because they want the job. So they're kind of forced to consume this technology, and they have no idea if it works or not or if it discriminates. And I have a lot of thoughts on sort of checking the effectiveness.

So I think what we need is, like, much more transparency and explainability. So if we could give even workers or job applicants or workers sort of the right to get their data back, that would also help us researchers sort of understand. Does this technology work on a grand scale?

Not on one person, but maybe on hundreds of people. I think what we've seen that does not work. And I've reported on this for MIT Technology Review. There is kind of third party audits where companies that build the technology pay third party auditors to check if the technology works.

And there have been two in the AI hiring space and both are really problematic. In fact, in one of the cases, the researchers told me while I was reporting the story that they found problematic things where marginalized group actually did not pass and were discriminated. And that was not it.

That was omitted from the final report because the company paid the researchers to do this. So I think we need, you know, I can think of like sort of the National Science Foundation or other places who, you know, NIST and others to sort of have these maybe grants or incentives for researchers that are not affiliated, not getting paid to, to actually check these tools. Thank you.

Yes, I would echo that.

Go ahead.

Okay. I would echo the point about transparency. I think that transparency and reporting are incredibly important. And the sad fact is that typically both the technology providers making the tools and the employers using the tools are very resistant to that kind of reporting. So that's essential.

But the counterpart to that is to say, okay, so we're generating a bunch of big data about exactly what's happening with these technologies. How do we make sense of that? There have to be human resources devoted to actually analyzing the information as it comes in.

We can use AI for that, too, but there has to be some people at the helm there, and we know from past practice that making a law but not thinking about how to monitor and enforce simply doesn't work. Let's not make that mistake this time.

Yeah. And then I just had. So we had...

I'm sorry. Can you get closer to the mic?

Sure. Sorry about that. So I think we have SB 1446, which talks about self checkout, safe staffing levels at self checkout, limiting number of items, requiring workers, requiring a notice to workers and the public when new consequential workplace technology is adopted.

For me personally, I think there needs to be a legal framework of liability for AI. Because is it the person running the model? Is it the company that owns the model that's being run? Is it the employer that deploys it? There needs to be a way to actually hold people responsible in a meaningful, monetary way for violating the law or putting people at risk.

And so that's something that I don't have necessarily the insights for, but I know that that needs to be something that's established, because otherwise people are going to flout the laws until there's actually a way to prosecute them and make sure that they actually pay in a meaningful way for it.

And if I may, I think that's actually a critically important point. And I think one of the things that I have said publicly many times is that we're at a moment in time, the precipice, I think, is what Sara said, where we have a choice to make. And when the Internet came to pass, we made a choice.

Well, not us in this room, but the government made a choice to pass Section 230 and ensure there was no liability for social media companies. And we have seen the results of that, and we are at a moment in time where we get to make a different choice.

And I actually think it's really critically important that we make a different choice because it is now really hard, and we try every day in this building to try to ensure that our children are protected from the harms of social media. And putting that back in the bottle is really, really challenging. And so I think the liability piece is critical.

Assembly Member Carrillo.

Thank you, Madam Chair, Madam Chairs. I just, one, I want to echo the comments that have already been made. Pero quería decirle unas palabras en español a la Señora Arciniega. Que bueno que estés aquí hablando. Los problemas y causas de su trabajo son algo muy importante que todos los trabajadores en todo el estado de California han visto y oído aquí en este lugar donde hacemos las leyes necesario para poder proteger tu trabajo y tu familia y quiero agradecerte por venir.

Creo que estamos en un momento muy oportuno. Lo que ha comentado el comité es que la tecnología de inteligencia artificial y la causa del daño que está causando a muchos. Para muchos trabajadores y empleos en general en todo el Estado es algo que tenemos que ver y es importante para el futuro de nuestras familias y nada más. Quiero decir gracias por venir.

I want to just, you know, I want to echo all the comments that have already been made, but I think, in particular, when it comes to employers and workers, we're at a critical time frame. I can, you know, balance, help balance the state budget, but I don't know how to count avocados on a self counter. Right.

And nor do I want to take the time to learn how to do that. And so I see elderly folks in my community struggle with buying groceries because there's no one to help them at the checkout. And that's certainly something that needs to be discussed, especially in communities where, in my community, for example, this was the second time that voters got a ballot to vote by mail, and still there's no education on it.

So if voters can't figure out how to vote by mail, can you imagine what it's like for them to try to check out their own groceries at a grocery store when we're already struggling in a variety of different issues? And so it's a challenge.

My work in the AI space has been focused on artificial intelligence being used in elections and how, you know, it's very possible for fake people to be created on a mailer, on a digital ad, voices that aren't really a real person, but you get a robocall, and you just don't know, like, if that's a real person.

And we currently have no disclaimers on what that could potentially look like or the influence that that has. And so we're working on policy that's moving forward in that space. But I think it's really difficult to put the genie back in the bottle.

And so now it's about regulation and how we get ahead of that, all these issues that are moving forward. So, again, incredible, just input. I think as the Legislature moves forward and tries to figure out, at the end of the day, how do we regulate something that's already out but do it in a way that's efficient. Right. And that at the end of the day, supports the vision of an informed state that can also inform the rest of the nation. So thank you for your work.

Oh, sorry. She just wanted to add something.

Yeah, please.

Gracias, Señorita Carrillo, por escuchar el testimonio.

She just wanted to say thank you, Assembly Member Carrillo, for listening to her testimony.

I just wanted to add one thing to, I think, the remarks that Assembly Member Irwin made about upskilling, which was the importance of those conversations between the workers and the employers. I also think it's really important that we engage in that as government, that we make sure that we are not falling asleep at the wheel in providing resources to ensure that people have the skills to thrive in the new economy.

And I think that it's really important that even during hard budget times, those are conversations that continue with our partners in labor and how we can be pushing that forward through our incredible institutions that are here in California.

On that note... Oh, go ahead.

Just one quick idea that, I think, you talked about how can we make sure this is efficient and it's working and bring the workers to the table. And I think they could be sort of co-decision making of companies and their workers that you could put into collective bargaining agreements that workers and the union has the same weight of say which technology will be used.

And that, first of all, it has to be disclosed that this technology is being used because that's a problem that doesn't happen. And then secondly, they have to agree that this particular technology is being used. And I think a good model might be. As I said, I'm originally from Germany, and there is a workers council that any company that is larger than five people, there's nothing to do with union. They can vote for a workers council.

And there's a law in Germany, it's a Section 86 in the law, that says any technology needs to be co-decided by the leadership of the company and the workers council, which is like a few people that are voted in by all of the workers in this company.

And I think you can see what's kind of interesting is the worst of this surveillance technology has been pushed out. You know, there is no individual monitoring of warehouse workers in most of Germany because the workers have said no. They said, you can, like, look at our teams, how productive we are, but we are not allowing that you track one of each of us and fire us because we are under like sort of some sort of productivity algorithm. Just an idea.

Thank you. I just, again, I want to quickly just kind of echo what my Chair, the other Chair, mentioned in terms of what we've been watching in Privacy, in the Privacy and Consumer Committee, which is a lot of promises from the technology work, you know, people that are doing this technology. And we like to talk about how we're the fifth largest economy in the world, and a lot of that technology that's being used is being created here in the State of California.

But unfortunately, I've seen the story up close and personal where we are made, a lot of promises are made, a lot of state dollars are used to enhance those promises. And in my district, or next to my district, we have Tesla, where Elon Musk promised us a new green jobs.

And what I've seen over and over is those workers coming to me and talking about the injustices that are happening in the plant and their inability to even purchase one of those Teslas that they are making every single day. And so, you know, what I don't want us to do is to continue to wait to see what happens or to depend on the promises that are being made to consumers, to us, to workers that they're going to be taken care of. And then we find out later that they're not.

Just like we're finding out in social media, where we hear every day the horror stories about what's happening to our youth, and then we hear the companies come in and promise us that they're going to fix it or that they're fixing it, and we don't see the results of that. So with that, I want to... Unless there's other questions? Okay. I want to pass it over to, I want to thank our panelists. Muchas gracias. And call in our third and final panel.

Thank you.

So our third panel we have speaking will be SAG-AFTRA a National Board Member and actor. There's been a lot of buzz about our next panelist, Sean Astin. Please, any chair. Andrew Gena, Research Director for the Amalgamated Transit Union, and Sandra Barreiro, government relations advocate with SEIU California.

Hi there.

Whenever you are ready. Yes, we can...

All right. Wow. First time here. Thank you, Chair Ortega and Vice Chair Flora, Members of the Assembly Labor and Consumer Protection and Privacy Committees. Thank you very much to my fellow participants, attendees, for the opportunity to share with you all today.

I want to acknowledge my SAG-AFTRA Union fellow members from the Los Angeles Local Government Affairs and Public Policy Committee, and the rest of our helpers, Shane Gusman, who you guys know. My name is Sean Astin, and I am a professional actor.

But my appearance here today is rooted in the fact that I was elected by the members of my union to serve on our National Board, representing 160,000 members nationwide, and on my Los Angeles Local Board, representing 80,000 members.

And if you add in our San Diego Local and our San Francisco Local from the State of California, about 90,000 members. As you can imagine, when it comes to artificial intelligence, all of our members are extremely concerned, and with good reason. We're already feeling the unsettling effects of this technology in our daily lives and on our livelihoods.

More than that, our dignity is under assault. I want to share two concepts that I think are both true. First, there are times throughout human history when technological advances change the very nature of our experience as human beings. Artificial intelligence is clearly, obviously just that.

10 years from now, the way we live will have changed in ways we can't even imagine right now. Two, sometimes products or systems come along that change consumer demand, rendering earlier technology obsolete. That is not what is happening here. This is not a buggy whip situation where they were made redundant by the advent of automobiles.

This is more akin to horse thievery or carjacking. Artificial intelligence is exciting. It's wonderful. These companies are driving, not just innovation, but jobs, our economy. We love our tech companies, but the unethical use of their innovations must be identified. This hearing has been an eye popping consideration of just some of the things that are happening.

The unethical use of these innovations have to be identified and stopped at all costs. Tech companies large and small, the studios, networks, and streamers, are not responding to market forces. Their technology adventurism is nurturing a systematic dismantling of our profession, where maintaining a career is nearly impossible for most of our people.

10% qualify for health insurance each year. 10% of that 160,000. Where once a performer could achieve a middle class existence and their share of the American dream, they are now unnecessarily forced to approach the industry as gig workers. SAG-AFTRA, along with the Writers Guild, IATSE, the Teamsters, and other sister entertainment trade unions, are the canaries in the coal mine. We are among the first to confront this particular threat, where, in the case of our performers, our identities are being siphoned away from us.

Our performers are the first to have our names, likenesses, voices, and body movements abused by this technology. But we are in the vanguard. Last year, our historic 118 day strike led the way in the labor movement. Our 118 day strike was our membership saying no.

They demanded that we do something, anything, to address this existential threat to our community. Some advocated for a total prohibition on artificial intelligence and digital replication. When this couldn't be achieved, we zeroed in on three core tenets of how our people are related and relating to this technology, some of which has been indicated. Consent, compensation, and usage.

The language we achieved in the contract was groundbreaking, as far as it went. But now, as we are inundated with developments and advancements that we knew were coming that couldn't yet be addressed in bargaining, we are existing in that awkward space of waiting for another two years in our contract cycle to play out, during which time the incredible speed of advancements won't be able to be challenged.

The average member is caught between understanding the real life implications of the agreement we reached and the contract enforcement, which feels pretty distant, as the companies test the boundaries of the good faith portions of the agreement in ways that, throughout American history, big business has often done with little care for the human beings they hire.

For example, per the contract, they're supposed to give our people three days notice before requiring a decision to be made regarding consenting to the digital replicas or use of artificial intelligence. They are supposed to give us a separate, clear, and conspicuous consent document with a reasonable description of how they intend to use it. Well, they are not honoring the spirit or the letter of that agreement in cases.

We have evidence of them requiring every performer to sign off without offering sometimes what their time, advance time that they're supposed to, but also without offering the information as to what they intend to do with it. You can imagine the complexity and the practicality of enforcing that rule.

So our members are asking, wait, if you're going to require my consent even when you have no expectation of using it, why are you doing that. They have reasonable questions about what sort of biometric data is being captured, how it's being stored, and to what extent they're using it beyond the spirit, and maybe even the letter of our agreement.

Our members are left to wonder what kinds of bias are happening in the hiring process. Our union is powerful, and we enforce the provisions of our contracts. But given the size and nature and scope of the challenge we have, we need your help. Only government has the ability to impose sufficient enough consequences in the only language these big companies understand, money. I'm not going to bring up specific legislation, though dozens of states and the federal government are managing a plethora of bills. What I would say about the dynamic I just described is that we need... Is that for me?

My wife pushed that button. I'm sure of it. A couple more pages or no? Is that okay? All right. What I would say about the dynamic I just described is that we need an informed consent language bill of rights, not just for us, but for all people.

This would be one tool to help prevent or deal with deepfakes that can terrorize anyone, including legislators, made to look as though you are behaving in ways you wouldn't behave or supporting positions you wouldn't support. I'll finish with what I believe the biggest AI issue is facing our industry and our members.

The tech companies make an audacious claim that doesn't pass the smell test. They assert that when digital information is ingested into their large language models, it all goes into a black box. The characters and information go through some magical, mystical, unknowable transmogrification, such that when it comes out on the other side, there's no trace of its digital origin. Our ability to police their usage is impossible.

Performers are left to scour the known universe and look with a magnifying glass to see how much of their noes, or what notes of their voices, or how a person or a creature moving is them. I believe, in this case, we are dealing with dishonest brokers.

We need the full weight of the law to level the playing field because, where the limits of bargaining end, only you can hold them to account. We must establish providence. We must protect the chain of title to our identities. Anything less and our people feel as though a piece of our soul has been taken. We feel violated.

Our identity is who we are. It's the vessel that holds our reputation and the emblem of our value. In 2005, the industry standard was to add watermarks to all filmed entertainment. In this way, they cut down on piracy. When it's in their interest to protect their products, they are tech geniuses.

When it comes to protecting our product, our very selves, they pretend they just don't know what to do. We demand that all data that contains our image, voice, likeness, movements, and other critical information unique to us, such as our union number, member number, be captured, embedded, and protected.

Whether it's through blockchain or other technology, we're entitled to know how we are being used. I sincerely thank you all very much for your laser focus and quick action on this profound issue of public import that is incredibly complex and often very dangerous. Thank you.

Thank you.

Happy to answer any questions that I'm capable of.

We will do that after we go through our next panelist, Andrew.

Thank you. Good afternoon, Chairs Ortega, Bauer-Kahan, and Members of the Committees. My name is Andrew Gena. I'm the Director of Strategic Research for the Amalgamated Transit Union. We are the largest labor union representing public transit and allied workers in the US and Canada.

We have almost 200,000 members, including transit, paratransit, school, and interstate bus operators, subway, light rail, and streetcar operators, mechanics, maintenance employees, and other related classifications. In the US, ATU members work in 44 states in the District of Columbia. Here in Sacramento, Local 256 represents bus, paratransit, microtransit, and light rail operators employed by the Regional Transit District.

ATU members play a vital, irreplaceable role in the delivery of public transit. They work diligently, day in and day out, ensuring that riders get where they need to be, work, school, the doctor, the grocery store, safely and on time. Our union ensures that jobs in the industry provide multi-decade careers with good wages and benefits, as well as opportunity, security, and dignity for transit workers. As with other industries, public transit faces an onrush of new technology.

Our task is to shape innovation such that it enhances the lives of our members and the communities in which they work and live. Some technologies, like advanced driver assistance systems, could make work safer and less stressful. However, these systems must be calibrated appropriately to provide accurate information that avoids cognitive overload of operators.

It is critical that manufacturers seek input from our members as the end users of these systems. Successful teaming of skilled humans and automated systems created an impressive safety record for aviation, and we think that approach could be adapted for public transit.

But in a bid to capture market share, corporations are pushing autonomous vehicle technology as far and as fast as they possibly can, with little regard for workers and the public. From 40 foot buses to microtransit vans, companies are trying to develop and market autonomous transit vehicles of all sizes without first ensuring their safety and effectiveness.

Companies make unsupported claims about potential cost savings through labor elimination, improved safety, and improved quality of service. Often, reality is less impressive and possibly more expensive. We've seen small shuttles that can seat six, travel about 12 mph, require onboard attendance, and remote vehicle operators.

We've seen microtransit vans that can only drive on specifically mapped streets and still require human operators under certain weather and road conditions. These are just a few examples. But in a scramble for public contracts, publicity, and market share, AV companies have thrust us into a real time experiment with unproven technology.

This raises concerns about job safety and quality of life for tens of thousands of transit workers across the country, as well as the safety of passengers, motorists, vulnerable road users, and pedestrians. A recent AAA survey found that two thirds of respondents were actually afraid of self driving vehicles, and for good reason.

Whether it's being involved in accidents, some of which have been fatal, dragging a pedestrian, crashing into a tree or a telephone pole, illegally driving in a bus lane, interfering with first responders, blocking intersections, or failing to identify children, autonomous vehicles and automated steering systems have shown that they are unpredictable and unsafe.

In contrast, public transportation is 10 times safer than traveling by car, according to the American Public Transportation Association. Self serving safety claims coming from AV companies should be treated with skepticism when we know that the safest transportation option is to simply take the bus.

While the AV industry's marketing might be attractive to cash strapped agencies facing worker shortages, transit leaders and lawmakers at all levels of government should responsibly use public money to support proven modes of mass transit. Rather than indulge in questionable AV projects, investments should be made in job quality to overcome the operator shortage and guarantee reliable service.

To that end, ATU has taken a leading role in developing joint labor management workforce development programs and registered apprenticeships to attract and retain workers, particularly right here in the State of California. As we grapple with the challenges presented by new technology, we are looking at state and federal legislation as one area of action.

Not surprisingly, we have met resistance from the industry, including the Autonomous Vehicle Industry Association, which is supported by Google, Amazon, General Motors, Ford, Uber, UPS, among others. Despite that opposition, we have had some success.

For example, ATU was happy to endorse California's AB 6, which was enacted last year, and that legislation requires transit agencies to address important issues through collective bargaining, including advanced notice of autonomous vehicle projects and transition plans for workers affected by new technology. AB 96 also requires bargaining over retraining and upskilling for workers, something we have so far not seen the industry take seriously.

This is a good starting point, but ATU's fundamental position is that, in the interest of safety, a properly licensed operator should always be on board a transit vehicle and engaged in its operation, regardless of automation level or size. We are eager to support legislation that states this outright.

In Maryland, for example, we supported legislation that would have required a human operator onboard any autonomous vehicle operated by or for a public entity providing public transit service. We hope to bring that legislation back up next session and pursue similar bills in other states.

Generally, we are working with allies across the country to support a variety of regulatory bills. Some examples include mandates for public data sharing and granting municipalities regulatory authority over AV deployments. We are also opposing bills that allow unregulated operation of autonomous vehicles on public roads.

At the federal level, we are partnering with Representative Chuy Garcia on a bill that would significantly restrict recipients of FTA funding from operating or contracting for the operation of autonomous vehicles, regardless of their size, until the Secretary of Transportation can establish a robust safety framework that includes vehicle capability testing, human machine interface standards, in vehicle safety operator training, and data transparency.

A standardized safety framework is essential for safe deployment of this technology and is currently being overlooked. ATU strongly supports federal regulation of autonomous vehicles and public transit to avoid a patchwork of differing regulations across the states. Ultimately, we are focused on the critical safety implications of unchecked AV deployments and the impact on good mass transit policy. Thank you for your time.

Thank you.

These workers are intimately familiar with state technology and databases because it is part of their daily lives. So first, I'd like to correct the stereotype that state workers or unions are anti-tech. If there's hesitancy around the Governor's AI Executive Order, it's based on state workers' firsthand experience with previous modernization projects that either failed or created significant disruptions to public service. So, for example, FI$Cal was launched in 2005. State agencies still struggle to use it.

The same goes for CalWIN, CalSAWS, and the EDD program that delayed benefits during the pandemic. When tech fails, it's the workers who suffer blame and the wrath of the frustrated consumer who feels like they've been given the runaround by an endlessly looping chatbot. So--and remember, just this year, 10,000 state worker positions were eliminated because of a budget deficit. At the same time, the Governor's executive order directs the state to explore a technology that purports to replace human workers.

So I think given this context and given the hundreds of millions of dollars spent on previous modernization projects, a healthy dose of skepticism is warranted. The goal should never be to replace the human element in government. If AI systems are intended to support rather than replace human workers, frontline workers must be considered as end-users.

And to succeed where previous modernization efforts have failed, the end-user needs to be incorporated from the earliest stages of adoption and remain engaged through testing, implementation, evaluation, and development of workforce training. It's frontline workers who should be answering basic questions like, 'do we need an AI solution here?'

'What problem are we trying to solve with this technology? Who will we serve, and what are the special needs of that population?' By answering these questions, workers might be able to identify alternatives, such as updating existing systems that are cheaper, cause fewer disruptions, and require less extensive workforce training.

And this determination alone could save the state millions of dollars and prevent disruptions to services. Conversations about workforce training right now tend to evolve around basic operations, but we want to make sure that there's a few other areas that deserve careful attention. I heard talk about upskilling, but I think we also want to look at preventing deskilling.

So deskilling occurs when a technology replaces a skill or function previously performed by a human, and the end result is lower wages. And we can take the examples of APGAs or program analysts, which are one of the largest classifications of state workers, and they serve in nearly every department.

One of their core functions is analysis, and it's in their job title. But if program analysts start relying on AI to perform their jobs, they're going to be--the analysis gets replaced with input and editing, and input and editing isn't as highly compensated as analysis. So that would be an example of deskilling, kind of near and immediate deskilling. Workers also need to be trained on the risks. Right now, I don't think that the trainings are going far enough to provide that information.

And we're all familiar now it's discrimination, privacy, and security, because the reality is that while GenAI can sound human, it frequently generates inaccurate or inappropriate outputs. So training needs to include how to monitor and take protective measures against these potential risks.

And then lastly, we need to make sure that workers are provided ongoing training that makes sure that they can still perform their job when systems fail. Remember, the stakes are much higher for government than private sector. If government systems fail, people could lose their jobs or their homes.

But despite concerns, there's still a lot of excitement over AI and the opportunity to do things differently. You know, AI could possibly help finally address chronic caseloads, issues that result in high burnout and turnover for state workers. Or maybe AI could free up capacity for state workers to focus on the core functions that really bring them into state service. And in the past, modernization projects haven't utilized the experience of frontline workers during the selection and design phases of modernization projects.

But now we have the opportunity to tap into the resource of millions of hours of worker experience and use it to improve the lives of California. I hope that you will consider our recommendations and consider us as equal partners as we move forward. Thank you.

Thank you. Assembly Member Bryan.

Offer for consideration accepted. No, I think what's interesting about this panel and the previous panel is that we tend to think of how innovations and technologies impact what we are working on or the people we represent.

And when you come to a space like this, you see that all of us share an interconnected struggle that hits us in a different, nuanced way, but the theme that I'm catching is that we're not against innovation. The state shouldn't be against innovation. We shouldn't be the leader in AI across the country.

But it's to support a workforce. It's to support our growing economy and industries, not to replace and supplant. And Sean, as you mentioned, in some industries, it's difficult to support, something that's so unique and proprietarily creative to a person. It's theft.

And I think that's something that's incredibly important for us to think about in both of these committees, how we wrap our hands around that, how we regulate this in a way that protects the authenticity of the input that's been given to us by the folks who are doing the real work. And so thank you all for testifying from each of your unique and independent perspectives, and also, I know we're waiting for those federal regulations, but it's also not lost on us that as California goes, the country goes.

And so whatever we can do to be a part of advancing this conversation in a way that's meaningful, this information is valuable, but lean on us and push us to do that, because that's what we should be doing.

Can I just say that they can afford it? And all labor knows that when management does the right thing, it's better for their bottom line, it's better for the community. So we want to help them be better, and you guys can help us in that.

We've got your back. And also, thank you for reminding me that I am not the only person here who's buzzer-trained by my wife.

Assembly Member Carrillo.

Thank you, Madam Chair. Sean, we all want to know where you live and who is like the member that represents your area and when they're terming out so that maybe you might consider running for office at some point--is that your area?

I don't want to tell you because that person will think that it's a threat. So, you know, I live all over. Calabasas is where I live, my wife and family.

I am your representative.

There you go. So it's no threat? Yeah, I will. I have. I'm sure.

Just opened up a whole other can of--

I know. Assembly Member Carrillo, please. Your question.

Thank you, Madam Chair. I serve. I have the distinct pleasure of serving on the California Film Commission and led the efforts to the last extension of the California Film Tax Credit Program.

And one of the challenges that we face as a state is runaway production and the reality of what that means to our economy, the impact that it has to jobs, great union jobs, but as well as local communities where filming takes place in California. And despite the challenges from various other states coming in with additional tax credit programs that we, the State of California is fighting against and hoping to save these jobs, now, the industry itself is facing challenges within the workers, the actors.

Voice over talent, a variety of different sectors in which artificial intelligence is playing a role in what that could potentially look like, but from an audience perspective as well, like what is it that we are receiving and what is truth and what is not and what is being manipulated with artificial intelligence.

I think, Madam Chair, you have a bill that addresses that issue as well, and so it's an interesting space because I believe that art is a reflection of our society and vice versa, right? And so what we consume, the products that are being created, says a lot about where we are, who we are, and how we get to move forward. And there's, it's difficult sometimes when you're up against industries and you're up against interests that have their bottom line, you know, at the forefront.

And so I think you coming up here, I want you to know, and your, you know, fellow members at SAG and other entertainment unions and the entertainment union coalition, you play a significant role in ensuring that we are making appropriate decisions from the creator perspective as well while we move policy forward.

And again, like to the other unions that are here as well, like the conversations with the members and how it's implemented from the bottom up, you take a bus, you're on transportation, you're working, you call EDD and you want a member of SEIU 1000 to be able to help you. It all matters. And I think we as a state are lacking in our technology. We saw that during pandemic where the EDD infrastructure just didn't work. And it took a long time for us to get back on track.

And so again, just want to thank all of you, and I think we're listening to some very real and honest testimony that can lead the Legislature forward in creating policies that at the end of the day support Californians.

Thank you so much.

Thank you. You know, I think what I, one of the things I love about this is I think we have two unions represented here who actually have used AI, right? Everyone thinks of AI as what came out a year ago, but AI is a lot older than that, and I know our state workers have been using it for a long time as it relates to call centers, and actors have used it in ways that have benefited them for a long time to cover them up in nudity scenes, et cetera.

And so I think it is an important point that AI is not even as new as people think, and that a lot of our workers have been adopters, users, beneficiaries for a long time, and so it is not as if we need to turn our back on AI, right? We need to learn how it has been beneficial, continue to use it in those ways, and not allow it to be introduced in ways that are hugely problematic to our workers and to consumers, again, as my colleague mentioned.

And I think that's really critically important. I wanted to ask--it's funny, as we talk about transit, I now live in the suburbs, but when I lived in San Francisco, I didn't have a car. I was a big bus rider. When we started having a conversation about there not being a driver or on a bus, I actually thought it was the craziest thing I'd ever heard because as a woman riding alone in San Francisco, I was often not safe on the bus, right?

And so the idea that I would be on there and there would not be a city worker was appalling to me. I think there was a discussion around school buses. I was like, I wouldn't put my kid on a school bus without a worker.

Now, again, it's an interesting idea on those San Francisco buses, if that person would have been positioned to help the consumers more, right, if we get to a place where the technology is such that that's able to do that, but the worker's still there, I think that's exciting. I'll be honest.

I was in situations where the driver couldn't turn their back on the road, and stuff was happening in the back of the bus that was really frightening--I'll be frank with you, right--and I think that's true for a lot of consumers in California. And I think we don't want to put our drivers in that position, but it's the reality of their jobs.

I mean, I think, I don't know if you want to mention that, but I do think when I put my kid on the school bus and my kid rides the school bus every day, I'm putting them in the hands of that driver, right? And I'm entrusting that driver to get my child to school safely, my young ten-year-old. So, you know, I do think that these folks carry roles beyond what even the technology anticipates, right, and I think that's critically important.

When the Waymo car got lit on fire earlier this year in the middle of Chinatown, I think it hit that point home. You can't really keep these vehicles safe when they're on their own from wherever those people are that are monitoring the cameras in those cars.

And so, you know, I think, as you mentioned, we have done some things to think about how to bring the voices of your workers to the table as we look at the future of transit. I represent a district where we actually, the reason I no longer have a bus rider is we don't really have bus service. It's actually a huge problem where I live. We have a huge last mile problem and we have no buses to take people to the BART stations.

And, you know, so I think that the future of transit, I hope it expands in communities like mine and other communities that have bus, but not enough service for the workers that work hours that are nontraditional, if you will. And as we look to a future with more transit, is this part of how we do it with workers at the table, I think is really critical, and I think also--so I want to sort of point that to you.

But also for SEIU workers, I have raised concerns previously about the EO myself because I think that I've been concerned about the privacy implications of bringing private companies into a sandbox with California consumers' data, data they trust us as government to hold in a different way than they trust private government to hold and utilize, and to ensure that as we engage in these practices, we're protecting California's consumers the way your workers do every day.

And in a way, I'm not sure that I trust the Amazons that want me to give away my children's health care data to do. So both of you, I think, engage in practices where AI is coming into play and how do you think we think about that in the engaging as a government to protect consumers and workers?

Quickly, I'll say you raise a point that I think goes to a little bit of the deskilling conversation that has happened today, where our members who operate buses earn a decent living, in large part because they have commercial driver's license, they have to pass medical exams, and they have to submit to drug and alcohol testing.

So all of that sort of combines into them being highly skilled operators who are very proud of what they do, literally driving the vehicle down the road. And a concern we have is if you take that away from them and you're left just being a supervisor of the public, that doesn't have the same skill level as being a bus operator does today. I think everyone here knows that as soon as labor can't command that income, management's not going to give it to you. So I think that goes to the deskilling, the displacement of work, and I think that's something we're very concerned about.

Can I ask a follow-up question? So is the concern--I think I'm hearing right, so I wanted to clarify--the concern is because there's been a lot of conversation in very different contexts, not AI with the Building Trades around some of our environmental work that is having a similar impact on--I mean, I represent an area next to their refineries.

So as those jobs have been impacted by our environmental moves. It isn't for those workers that I've spoken directly to as much about it is about wages, benefits, and all the things that come with being a union worker, but it's also about the work, right? So what I heard you say was there concern is the wages, benefits, all the things you can command. Is it also the work? I just wanted to address that because that's not what I heard you say.

Yeah, I don't want to downplay the personal connection. So we heard stories before from the brother from UFCW talking about the connection between customers and the cashiers. Our bus operators have the same relationships with the people that they drive.

I miss my bus driver.

So they do have a personal connection with the people that they drive. They're part of their communities, and that's extremely important as well, so I don't mean to downplay that at all. I mean, I think, you know, to--I guess to close with how we started thinking about the dystopian option, right, we're talking about a society where people ride in vehicles that don't have a driver or other humans in them.

They go to the store and don't interact with a cashier. They spend their time online, on Facebook, interacting with things that may or may not be actual humans, bots. So, you know, it adds up to something that is not a society I would want to be living in, and I think that's the human connection is certainly important.

Yeah. Thank you.

I have a thought about what you said about our business. So we have our wages and working conditions process where any member can come in leading up to a negotiation and their concerns inform the plenary and our proposals that we push across the table.

When this technology came in, you know, 25 years ago or more, it was so exciting. It was so exciting. It was a new frontier. It was, what are--how are performers going to give a, deliver performance with the dots all over their head or what's it going to be like to animate a creature performance?

What we're in now, and it's hard for a lot of people kind of in the public to understand, and it becomes a political football, is the idea of the entertainment industry as an actual industry. We have actual workers, workers who work. It's like a factory job. And when you end up in a situation where the industry is being tightened, there's a natural contraction happening anyway.

But when the attitudes and the behavior and the hiring and the demands that are placed on you end up creating a kind of punishing experience for those performers, that's when you need good bargaining and it's when you need good legislation to back that up. Yeah.

Yeah. And I'll say it's interesting, in my work on, you know, name, likeness, and image, one of the things I've seen is the distinction between the studios, the movie studios, and the Recording Artists Association and how they really--the relationship with recording artists, with their industry is very different, right?

And there's sort of mutual benefit in protecting the intellectual property of their artists in a way that I think has been really fun to watch, to be honest, because they've aligned the interests such that they believe protecting the artists isn't their benefit. And to your point earlier about, we know that when business puts workers first, it helps the bottom line, and I wish all industries would get there.

I want to acknowledge a guy who's here, Erik Passoja, who is our member. He's actually the Co-Chair of the Los Angeles New Technology Committee, and the kinds of work they're doing in their Committee is just mind-expanding.

And so when you talk about, like, how technologies are being woven and used and shaped and where we can balance from each other, it's an exciting challenge. We don't hate tech, just want to protect our workers.

Yeah.

And that, I will pass it over to Assembly Member Irwin.

Thank you very much, and I think this more for my constituents from Calabasas, not your constituent, because when you're talking, when you're talking about transportation or state workers, you're talking about within state boundaries.

So you're talking about workers that are in California, obviously, in the--if you're talking about the movie industry or the recording industry that's going on throughout the United States. I'm doing a bill right now on transparency and training data, so trying to really have these companies talk about who they trained on, and hopefully, you know, getting--eventually that will lead to--

That's so important. I'm so glad you're doing that.

Well, we will see where it goes. But this is, and this would be a bill that only affects work in California. So the question is, for work that spans across the U.S., how do you see that, how do you see that playing out? There are a lot of states that are not necessarily as sympathetic to union workers. They're looking at just the bottom line.

That's a great question.

And there is a lot of discussion about we need the feds to do something, but all of us know quite clearly that issues like privacy, the first bill in California passed in 2018 and Congress still has not moved on that issue, so how would you answer that?

How do we get other states to make sure that they are playing by the same rules? Because eventually we are going to be putting the right type of guardrails and transparency on California will be doing it first, but that's not going to be the case in other states and we don't want, for instance, movies to move, you know, these companies to move to...Georgia.

Well, I studied a little bit about the right of publicity as a area of intellectual property law. And there, there's all of these--I picture it like a, like a battlefield, and you're going to move this piece into play and you're going to move that piece into play. We are a national union.

We have performers in New York. We have performers in Atlanta and Dallas and Florida. I mean, we're a national union, so the cool thing is that our contract travels with us, but in terms of legislation, it's California. You, I mean, we're working federally. You know, we will find a way. I mean, we have bandwidth issues.

You know, how can we work with, you know, we'll pick the states. Indiana has a great piece of legislation or a law, a great law that deals with this publicity right. So, you know, maybe we'll, our union will figure out how to dovetail with them.

But when it gets right down to it, you know, this is, this room that you people, that you're the ones who are going to, like, lead the way on this, that--so thank you ahead of time. And we will be making sure that we help facilitate the conversations with the federal legislators and whoever the hell else we need to include in the conversation.

The most important thing for me for this moment is that there is--and it's clear that there already is--actually, I was at a--Darrell Issa had a hearing, this similar type of hearing in LA, and he had Adam Schiff and Matt Gaetz in there and everybody in between.

And they were jumping up to agree with each other on this issue. And so then we did, I did a roundtable, and it had all the, what do you call them, the general counsels of a bunch of the studios. And they were listening very closely because they're ready to jump up and protect their First Amendment Rights.

But they also know that their entire libraries are in jeopardy. So they want to make sure that they're more than happy to use us as a blocker and kind of to run behind and achieve some of this stuff for them. So it's a great environment, I would think, to be legislating in, because aside from the all important tech industry that supports this entire city, you've got a lot of support across the state and around the country.

It's my last term, so, yeah.

You clearly have some fans here. You know, before I go to public comment, I just, you know, this whole hearing has been so interesting and eye-opening, especially after sitting and hearing after hearing, and hearing from the companies that are creating the technology and telling us over and over in some cases that they don't have the money to implement some of the ideas that we come up with or that, or the regulations that we're proposing.

And so they're often opposing many of the things that we're bringing to the table, but none of that has stopped us, including the reason I had this hearing today in the first place was to make sure that we are making, that workers are at the forefront, not the after fact. Technology should be working for all of us, not just the wealthy.

And it is up to us as legislators to continue talking about regulation and safeguards, regardless of the opposition because as was mentioned earlier, we've seen what inaction does when it comes to technology and social media and other things, and they haven't been to the benefit of consumers or workers.

And now we're at a pivotal moment where we do have a lot of the research, a lot of the experts who are, you know, giving us ideas and proposals as to figure out that balance where we can have technology. We love tech--I can't live without my cell phone.

We love technology, but we have to make sure we have that balance, and it's up to us, and one other point I want to make is that, you know, a lot of conversations about collective bargaining agreements and language in those are great, and we should have that.

But we also have millions of workers across the country do not have a labor union, do not have collective bargaining agreements, and therefore rely on us to come up with those safeguards. So on that note, I want to open it up for public comment. One minute each, and then we'll close this hearing.

Good afternoon. Ivan Fernández with the California Federation of Labor Unions. I'm here today to speak on behalf of Jillian Arnold, who is the Chair of the IATSE Artificial Intelligence and Technology Basic Agreement Bargaining Subcommittee, and also who is the President of IATSE Local 695, representing the California IATSE Council.

For reference, the California IATSE Council represents over 55,000 women and men across the state who work in film, television, and live events. AI was a top priority of IATSE's 2024 basic agreement with the AMPTP, which is why the agreement included substantial AI language.

IATSE's contract now stipulates that when an employer assigns the use of AI systems to an employee, that work is considered covered by the agreement. Moreover, no employee is required to provide AI prompts in any manner that would result in the displacement of any covered employee.

Per the agreement, any request to scan an employee must be cleared through a separately signed statement or writing, and in a first, IATSE's contract stipulates that consent to scanning cannot be a condition of employment.

With some limited exceptions, the employer will be required to negotiate with IATSE over any impact of the use of the AI systems may have on their members. Thank you to the Chairs and Members of the Committee. IATSE looks forward to working with you all to protect working men and women in California. Thank you.

Thank you, Madam Chair, Members, and staff. Mitch Steiger with CFT, a Union of Educators and Classified Professionals. Appreciate the opportunity to speak today to this issue and this technology that has now kind of come out of nowhere and is now everywhere in the education world.

It's reached a point where it can be used to analyze text, develop a curriculum, develop a slide deck, and essay assignments that can then be assigned to students who can then use generative AI to write their essays. They can then hand that back in, and the original AI can then thoroughly grade those essays.

So we now have technology where it's basically just two forms of AI communicating with one another in the education system. That's not usually how it's being used, but there are pieces of it that are being used that way. There are students using generative AI to write their college essays. A pretty significant percentage of them are doing that.

And it's easy to imagine a world where all of this technology exists and coexists well with teachers and students and it's being used in a smart way with strict limits, where it makes sense, where workers are involved, and where it protects both communities. We don't live in that world right now. We're taking the steps towards getting there and we look forward to being involved in the discussion to make sure that those regulations that will get us there take place. Thank you.

Good afternoon, Madam Chair and staff. Janice O'Malley with AFSCME. Just really appreciate the committee having this hearing and I think you can tell by just the breadth of the different workers within different sectors how technology and AI is really affecting the workplace, and to your point about workers needing to be within the forefront of this emerging technology, we also have to be at the table.

We've seen technology being used to create efficiencies in the workplace at the tune of millions of dollars and if that technology fails or doesn't end up working for them, that means that their employers has shelled out a lot of money that could have been used towards wages and benefits and other really important things that haven't been, that are at a loss now. And so, thank you so much for having this hearing. Look forward to having ongoing conversations on this.

Hi. My name is Erik Passoja. I am a professional actor and 25-year member of SAG-AFTRA from Los Angeles. In 2014, Activision hired me to do performance captor as a Belgian geneticist named Doc in the blockbuster video game Call of Advanced Warfare. After the game came out, a friend called and said, 'hey, Erik, my son just shot you.' Turns out Activision had also put my face on a different character that you can play for monthly fee to fight other players.

If your kids played this game, my likeness was shooting at your children, and they would shoot me, blow me off a cliff without my informed consent or even additional compensation, and that year, Call of Duty grossed more than Taylor Swift's album, 1989, Disney's Frozen and Guardians of the Galaxy combined. All I received was my original session fee.

And ten-years later, at the dawn of AI, the gaming companies refused to change, and so our union was forced to strike. We performers are uniquely positioned because our digital identity is our only business. You take away a carpenter's voice, they can still build, but you take away a voiceover's voice, they will never work again.

And the solution is informed consent and digital provenance, as Sean Astin has said. Pro-labor AI legislation is crucial, and it must be enforced, and we need your help to protect labor rights and safeguard the public from deepfakes. Thank you.

Madam Chairs, and I don't know if you're all the Chairs or who the Committee folks are and the staff and everyone here in the room, thank you so much for this opportunity to speak. My name is Susie Duff, and I am about to be a one-half of a Century member, very proud member of the Screen Actors Guild, and every year have earned my living in that capacity. And I'm here to offer you a resource. I have a tremendous gift to you in the name of a film composer, Richard Gibbs.

I was on the phone with him. He lives in my community of Malibu, California, as I was en route here last night. Richard is also a Governor at the Academy of Motion Picture Arts and Sciences, so he walks both worlds, and for about a decade, he has run a weekly session called the Composers Breakfast Club. They are dealing with this issue in and out, up and down, and he was explaining to me his ideas last night on the phone, and I felt like a little girl in her mom's high heels. I just wanted to go get something to eat.

And I said to him, 'if I get a chance, can I recommend you to this committee?' And he said, 'go ahead.' His name is Richard Gibbs, and if you'd like his information, I'll be happy to get it to you. He would be a tremendous resource, and bless you, and thank you for all the great work you're doing, and thank you, everybody, for being here. Thank you.

Thank you.

Yes.

Okay. Seeing no other public comments, I want to once again thank all of our panelists, especially the workers who traveled here very, you know, from a long ways to make it to today's hearing and make sure that your voices are heard and part of the conversation. [Comments in Spanish.] Thanks again. With that, we are now adjourned.