Senate Select Committee on Select Committee on Economic Development and Technological Innovation

Thank you.

I will call this hearing to order. This is a really great privilege to convene the first of its kind Select Committee that intends to examine artificial intelligence impact on the California labor market.

There's no question that the daily reports on this technology as a disruptor and early anecdotal data suggest both a change in hiring practices and an impact on the workforce. That is something that this body, this Legislature, is going to have to contend with and be on top of for many years to come.

I'll say more at the end of my opening remarks, but I just want to thank my colleague, Senator McNerney, who's been a tremendous leader at the federal and state level, to be here with us today. Today we're not just talking about technology.

We're talking about a force that is already reshaping people's livelihoods and with it the trajectory of California's economic development. Having served for seven years on California's Workforce Development Board, I firsthand experience prioritizing training as well as technology that puts California's workforce on the cutting edge.

California, and in large part my district, SD13, is the epicenter of AI innovation. And with that comes responsibility and opportunity. I would say too, to help do this responsibly. Last year alone, our state added more than 100,000 AI related job postings, more than any other state in the nation.

Over 15% of all AI jobs nationwide are based here in California. That makes us uniquely positioned to lead not only in developing AI, but in ensuring that its benefits are felt across our economy. From our major industries to our local communities and to all residents.

AI has the potential to increase productivity, improve job quality, and even create new categories of work. These are opportunities that, if harnessed correctly, can strengthen California's competitiveness and broaden economic growth. But alongside these opportunities are very real risks. In 2023, there are over 340,000. Just as an example, there are over 340,000 commercially licensed drivers in California.

And these are potential. These jobs that are real, very real jobs right now are potentially ones that could be lost to automation. Nationally, there are 2.2 million heavy freight jobs and 1.7 million light duty delivery jobs. Ups alone, it's estimated that around 330,000 warehouse and driving jobs could be impacted by automation.

These are good jobs with good wages and good benefits. And as I'll mention in a minute, we now have reports of hundreds of thousands of workers could be subject to automation at Amazon alone.

Additionally, bias in these systems some of these Legislature has worked on can shrink opportunities for those who would benefit most from them within the workplace. AI monitoring has already been adopted by a growing number of firms, raising concerns that technology could undermine trust, which is essential to a healthy economy.

I was proud to co author one Bill around workplace surveillance this year by Senator Ellore that did not make it through, but continue to work on this. Constituents see these risks.

A majority in California now hold an unfavorable view of AI, underscoring the urgency of building sensible guardrails that allow our economy to grow while ensuring that growth is inclusive and fair. If we do nothing, AI could widen inequality, destabilize labor markets, and erode confidence in both our economy and our government.

But if we act thoughtfully, we can ensure AI adopts adoption. Or we can try to make sure AI adoption lifts workers up, not leaves them behind. That means investing in training and workforce development, safeguarding worker rights, and ensuring transparency and accountability in AI systems. These are not just labor policies. These are economic development priorities and strategies.

Proud of some of the steps we've already taken to put some guardrails around AI. I was proud to author the AI Transparency act to add transparency to AI generated content that will take effect next year. Co author this year's joint author, this year's Companion Chatbot Bill, SB243 with my colleague Senator Padilla.

And again, this is something the Legislature has already been grappling with for a number of years.

I was proud to floor manage in the Senate, Then Assemblymember Gonzales's AB 701 which prevented warehouse operators from using algorithmic quotas that put workers health and safety at risk and worked on other bills to Promote Earn Learn Job Training act, opened doors for veterans, the unhoused, formerly incarcerated, to enter the workforce.

Employees talk ownership plans, other things I think they can be part of. If more workers are owners, they can make sure this technology is implemented responsibly. At its core, this hearing is about ensuring the future of California's economy, an economy that continues to lead in innovation, but also leads in fairness and opportunity.

I want to provide an opportunity for Senator McNerney to make some opening remarks. I want to thank him for joining us today. Thank him for his leadership in Congress before joining us here in the state.

And he's had a Bill this year also working in the space to again try to mitigate the impact of AI in the workplace. I look forward to then hearing from our esteemed witnesses how we can achieve these goals and have a start, really the dialogue in this body.

With that, turn it over to Senator McNerney, would you like to make any opening remarks.

I believe I will. Chairman, thank you for the opportunity. And thank you for inviting me and for holding this hearing. As the Chair mentioned, I worked in Congress as the chair of the AI Caucus.

We passed the AI and Government act, which basically sets up a center of excellence on AI in the Federal Government so that the federal agencies have a resource to draw upon when questions on artificial intelligence pop up.

And my focus with AI is to make sure and ensure that AI is safe and beneficial for this country and especially for California and for my constituents. As the Senator mentioned, there's a lot of really important benefits to artificial intelligence, and we're seeing that in medical advances. We're seeing it, and transportation and agriculture.

A lot of ways that AI is going to really improve our lives. But if AI is going to be safe and beneficial, one of the things I am most concerned about is displacements that would be caused by artificial intelligence.

And I think it's in everyone's interest to ensure that that doesn't happen, because if there's a large numbers of displacements, then there will be social unrest and that will cause political instability and so on. And I think it's in everyone's interest to keep that, to keep the lid on that, on that bottle.

So we're going to continue to do that as much as we can from the state here. And the great thing about being in California, as opposed to the Federal Government, you can actually do things here. And not only that, but California does lead the country.

If we enact good laws, the people in Washington look at the states as the laboratory for laws and for ideas. And California is the place for those things to happen. I'm glad to be a part of it. This year I authored with the California Labor Federation SB7, which we called the no Robo Bosses Act.

And the intent was to keep a human being in the loop. And in crude terms, if you're going to eliminate an HR Department, you ought to have at least one or two human beings to overlook the decisions that are being made by these automated systems.

But there's all kinds of other applications where we want to keep humans in the loop, whether it's critical infrastructure so that we don't see hallucinating and causing grave damage and grave harm to people and to infrastructure.

I think one of the ways to do that was exemplified in SB813, which we authored last year to create standards committees for AI. Different aspects of AI, whether it's data, whether it's earmarking, whether it's healthcare.

We need to have some way to identify whether an AI product is something that's safe and beneficial for people to use and implement. So that is basically my tent in this issue. And as the Chair mentioned, it's not going to go away.

It's going to be more important as we move forward because as we will discuss, there's big layoffs planned and there's big changes in the air. So we need to get this right as soon as possible.

And I hope a discussion, an open and fair discussion of the issues will help clarify some that we need to be accomplishing here in Sacramento. So with that, I'm going to turn it back to the chair and look forward to our discussion. Thank you.

Good. Thank you for that. And as you say, this technology is rapidly evolving. And so we will certainly this is going to be something throughout our tenures here. And. Yeah. And the importance. Thank you for. And the importance of California to lead here. So thank you.

I also want to thank Ian Higgins for my team who helped put together today's hearing. Thank you. And with that, I'd like to invite up Sarah Flocks. Sarah is a Legislative Director for the California Labor Federation. And our plan is we have three witnesses.

We're going to do one at a time and we'll do plenty of questions for each one following their testimony.

Mr. Chair, Senator. Sarah Flocks, the Legislative Director for the California Federation of Labor Unions. We represent 2.3 million union members across the State of California. And I want to start off echoing something that you said, Mr. Chair. Technology is not inherently beneficial or risky. In fact, unions have bargained over and adapted to technological change for decades.

And we have negotiated so that the benefits of technology accrue to workers and make work safer and more productive. The difference that we're seeing with the AI tools that are out there today is the speed, scope and pervasiveness across all, all industries.

These are tools that are specifically designed and marketed to employers to increase corporate power and profit by reducing labor costs and increasing managerial control. In fact, OpenAI on their website talks about how their tools can do work 100 times faster and 100 times cheaper than a human worker. That's how they are marketing their tools.

And this is not a future problem. They this is here now. A recent survey of employers found that a majority of managers who use AI tools use it to determine raises. 78% of them, 77 use it to determine promotions. And 66% of them use AI to determine layoffs. 20% let AI make final decisions without any human input.

This is what we're seeing is that without the guardrails and regulations from the government, AI poses an existential threat to workers, to our economy and to the social fabric that is already badly fraying.

So to better understand the landscape and what we're seeing in the workplace, I want to say that there's kind of two broad categories of the technology that we see out there. Although there are hundreds, probably of thousands, thousands of tools that are marketed to employers. The first bucket is labor replacing or what could be called automation.

And so we're already seeing that. I've seen it for decades. This would be self checkout smart carts Amazon go where you just walk out driverless cars. AI actors which we recently saw animators, scriptwriters, journalists, mental health chatbots replacing therapists.

All AI schools like the Alpha School that's opening in Folsom and just cutting off the entry point at any knowledge and creative industry jobs. There was actually a recent Stanford study. I believe Yale and HSBC have found the same things.

Is that these AI tools are really impacting young college grads because all of these entry level jobs are being automated away. It is just literally cutting the bottom rungs off the career ladder and which can cause a crisis. Then labor exploiting. This is something that's harder to identify.

But it is starting to be that you are more likely to be managed by a machine than replace. This is the very insidious part of AI that Senator McNerney's Bill tried to address. These robo bosses. This is algorithmic management which is fueled by mass surveillance.

These are some military grade tools that employers have access to to be able to control and surveil their workplace. This is Taylorism and scientific management and taken to an extreme run by robots first. We saw it in gig work. So Uber, Lyft and Amazon Flex.

These are drivers that are pretty much hired, managed and fired completely by algorithm with no human contact. We've seen drivers who have been fired for one bad customer review which falls disproportionately on black drivers who are deactivated at a much higher rate than white drivers.

You also see an Amazon Flex where you might take a different route or not be following the exact directions of the algorithm because it's raining or a road's closed. You are deactivated and cannot get a human on the phone to be able to talk to Shift key.

This is a nursing gig app where workers have to bid on shifts and the algorithm finds who will take the lowest shift so that they can pay them less per hour in warehouses. We're seeing mass management by algorithm.

In fact, one Staten Island warehouse, there were 13,000 disciplinary warnings sent in a year where workers after six would just be automatically fired. It doesn't matter if they're injured, if they're sick, if they're pregnant, if they're not meeting their quotas, they are being fired. They also have wearable devices. This is huge haptic arm bands that will buzz.

So you move exactly in the most efficient way to pick packages. This is one of the reasons Amazon has injury rates hundreds of times higher than other warehouses. We're also seeing this with white collar workers. There was a New York Times article in 2003 that went through this with tools like WorkSmart, Time, Dr.

Occupy, and the very aptly named Controlio, which is designed to fully surveil and control workers. They monitor keystrokes, mouse clicks. They have cameras that do retinal scans. They can see when you look away. It makes it so it's hard for a worker to even get up and go to the bathroom without justifying it to, to an algorithm.

United Healthcare uses these bots. It marks idle time. And so there were social workers and therapists who were complaining that anytime they talked to a patient and weren't looking at their screen, they were dinged for having idle time, even if they were counseling somebody with serious mental health issues.

Social workers were rated from 1 to 5 based on idle time and their compensation was then decided. So there's real consequences for these Control Cogito and bochi Vocai. I don't know how it's pronounced. These are tools that are used by call center workers across the board, whether people are providing benefits or taking, you know, insurance claims.

It is a algorithm that monitors your voice and probably can be hooked up for biometrics and then gives you clues. Coffee cup, you need more energy. Heart, you need more empathy. Speed up your talking, slow down your talking. Pay attention to the script.

They said that if someone were crying, you're supposed to say I'm sorry, even if they were crying with laughter because they just had the birth of a child. And so these are how algorithms are moving into the workspace. These are just a few of the examples. Another one has impacts on patients.

There is a nursing home called Brookdale. There's been a number of class action lawsuits that their AI, they had an algorithmic management, didn't take into account the needs of elderly and patients with dementia and so only allotted 20 minutes. And so they were desperately understaffed.

And so a number of patients have died because of the lack of care. So what are the harms to workers? You went through some of them. One is just bias and discrimination. Even though this is technology, human bias can be built into the system.

We see this with the impact in terms especially hiring, but in deactivations that drivers of color, especially with Amazon and Lyft that we've seen in some of the surveys are disproportionately deactivated more frequently probably because of customer reviews. Without human oversight. With human oversight you might be able to identify this loss of autonomy and deskilling.

These take away any thought, any compassion, any human expertise that workers have, especially in call centers or other care jobs. And also work speed up and intensification.

We've seen this in the number of injuries in warehouses, but it's happening across the board whether you're a therapist or even in the New York Times articles chaplains, where they were speeding up the amount of time that chaplains were able to spend with patients, their parishioners. And the other is suppression of wages.

We see with the wage bidding, with deskilling, with taking away a lot of some of the skilled jobs, the thought jobs that you're able to pay less and then also just invasive surveillance and then union busting. We also see this in technology that's able to be predictive or infer behavior.

I believe Amazon Perceptics is a company that Amazon uses and, and they have a union.

I guess it's like union openness index where you can rate workers based on things they just pull from the Internet, inferences from their behavior to say yeah, that person is definitely going to join a union and so we better keep an eye on them.

So in terms of recommendations, the first one we would say is that workers need to be involved in the development of this test technology because right now the tools that are being developed are ones that you can market to employers so that they can save money and increase productivity. That is not necessarily what's good for workers.

At the second level, have workers be involved in the procurement and deployment of the technology early. There was a Bill that was done, Mr. Kalra did, but it's done been done nationally to make sure that unions get notification 180 days before an employer puts out a bid for AI enabled tools.

This allowed for the workers to have input into what is actually needed, what would actually make the work safer and more productive and then the ability to negotiate over how technology is deployed.

Often workers know how workflow goes a lot better than managers, but managers are the ones buying the technology which is deesigned by people who have no idea what's happening in the workplace. So we need to kind of change that flow. And then of course, human oversight, this should just be a no brainer.

We should not have robots making these consequential decisions either about workers livelihoods or the people that they are assisting. Whether it's Medi Cal benefits or any other very sensitive issue. We have to have that not human in the loop human oversight where humans are able to make the last and final decision according to their own expertise.

And I know that you both and the Senate and the Assembly have done a ton of work and we look forward to working with you in the future.

Well, thank you for that. And you really lay out a lot of the risks. And you know, some of these things I've heard before, some I had not heard before. So I appreciate you educating us on what effect this is already having in the workforce and things we have to be concerned about.

And I know we'll discuss this later too, around college grads. I have two kids in college, so the impact on young college grads is very personal to me. And just sort of for your last point, I did do a Bill on banning health care companies from using algorithms in denials of care and proud of that.

But sounds like from talks about other folks I think we'll hear from later too that also in decisions around nursing and other things that we're seeing the impact. I have a set of questions I'm going to turn over to Senator McNerney first, let you lead off with some questions and then I'll have some.

Well, thank you. Thank you Sarah for your comments. Somewhat sobering. And of course that's why we're working together to try and find solutions, language that is acceptable to both workers and management. And I think that's the key. If one side is going to be totally blocking and resistant, we're not going to get anywhere.

We need to work across that aisle. Not across the Democratic and Republican aisle necessarily, but across the worker and management aisle. So some studies, including from Stanford, show that AI hiring systems can filter out the better candidates.

Is there some way to make sure that these hiring algorithms are fair and lead to the best outcomes, including bringing in the best candidates?

Yeah, it seems like hiring is where a lot of these tools are being used. I know there's been a number of bills that have been run about making sure there are independent third party audits for bias and discrimination in these tools.

I believe in that same Stanford study or there was another study that shows there's Also an issue when companies use the same hiring software, when there's a monopoly or just one, because it's filtering out the same candidate so they don't have any ability to access or to have jobs when it's just one algorithm.

So I do think it's very important that there's independent audits, but that also these tools are opened up and there's a layperson's description of what goes into it, because that's one of the biggest problems is one, workers or job candidates don't know what tools are being used and, and they don't know any of the criteria or the data that is fed into it.

I do think hiring, because it's so pervasive, it's one of the areas that should be tackled on its own because. It is. The opportunity for bias based on things that we don't even know is so rampant.

And because of the sheer power and speed of AI tools, there is no human to kind of see like, zero, these are the patterns and these are the people being rejected.

Well, I like your distinction that I hadn't thought through clearly between human in the loop and human oversight. So we'll keep that in mind as we move forward. I appreciate that. How can we make sure that companies don't just have hand off these decisions to algorithms without any human oversight or humans in the loop, either one.

I mean, you had an excellent Bill that would mandate that. I think that should just be kind of a standard that there has to be human oversight for the most consequential decisions. For us it's in the workplace.

But this is another areas of people's lives where there needs to be some ability, some real ability separate from confirmation bias of a reviewer of this information. And that I think it's not something that. And some companies do it voluntarily, but there needs to be a standard that this is just what happens.

Well, we worked hard on SB7. To me it was a no brainer, but there was pretty significant opposition. What do you think was the basis of that opposition?

I think for a lot of companies, especially in, I mean platform and gig and warehouses, they have managed to outsource all management to these algorithms, especially when you have a disposable workforce. And this is one of the reasons we're very concerned about AI is because it enables the growth and spread of a insecure, contingent workforce.

And so we don't want to see the gigafication, the uberization of industries that have traditionally not been. That's actually the title of an excellent report from the Roosevelt industry called the uberization of nursing. And that is a big fear.

Like we have to make sure that especially in industries where you're caring for patients and the most vulnerable, that we are not turning this into something where you just need a ready, disposable supply of workers that you can just deactivate fire if they deviate from whatever the algorithm tells them to do.

So you're telling me that corporations feel workers are disposable? I'm saying that in jest. I'm sorry.

Not all. Not all, of course.

So the other. Going back to my first question about how do we make sure that companies do keep, and I think workers, people in the loop somehow, what I got was a diversity of hiring tools.

So if all companies are using tools created by a single company or a single tool, then there's going to be uniform bias in applying that tool. And that would hurt a certain class of people one way or another. So that would be one thing to have a sort of diversity of tools and then some audit.

I'm just throwing ideas out to ensure that company A is using one or two tools and company B is using a different set of tools. And that way at least people have a chance if there's significant bias, which there will be. Comment.

I think that is a very good idea. I mean, the other idea is to have. I'm not saying that it's publicly funded, but we have these excellent universities. We have so many resources in California.

Going back to the development question, having perhaps a non privately controlled company, having a University that's able to develop some of these tools, maybe in conjunction with unions or with workers or with the people who are going to be using this tool.

And so in that development, not only do you have a larger range of a diversity of tools that employers can use, but they are geared to be more beneficial to the user. So the impact on the workers would be more beneficial, safer work, more productivity. And then hopefully the same would be for employers.

And so maybe there's a space for funding to have AI tools that are developed by the public for the public.

Very good. So are you seeing. I mean, one of the things I mentioned earlier was concern about the social impacts, the instabilities and political upheavals. Are you seeing the start? Are you seeing that happening yet? Because of. Specifically because of tech layoffs, Tech layoffs and replacements.

Two things. One, I think we can see the anxiety in the polling that is out there. In all of the polls that we've seen, there's a lot of fear of work by Work of workers, of AI replacing them, of fear and kind of anxiety caused by all of the surveillance, concern about privacy and the data collection.

I mean, what is fueling all of this is worker and personal data. And then there's just the lessons that we learned during deindustrialization and nafta.

When you have massive layoffs and destruction of entire industries that aren't replaced by anything, then you are left with a fraying social fabric or with mass unemployment, with an opioid crisis, with what is in the Rust Belt, except this is happening across all industries all at once. And so we are very concerned.

It's the reason the AFL has started a national working group. Our President, Lorena Gonzales is one of the co chairs along with Massachusetts. But they are trying to tackle this at every level because we are very concerned about the impacts, about the social impacts, the economic and the workplace.

Okay, I'm going to turn this back over to the Chair. Thank you, Mr. Chairman. Thank you.

Yeah, I'll add a couple questions and again, thank you. Just give me some time to reflect as well that, you know, my experience is that companies succeed when they're empowering workers. Right.

And I think in my experience also in technology that companies have been successful in startups are really about empowering workers and involving workers in all level of decision making. And so the kind of management by algorithm that you talk about is very concerning.

One thing that comes to mind just for us to, you know, as we continue on in this effort, you know, we can, in addition to legislation, we also have the tools of these kinds of hearings or we can bring some of these hiring tool companies in front of us and ask them to explain how these tools are being used and ask them questions.

Or when you mention if gig companies are deactivating certain types of workers, certain ethnicities more than others, we can bring them in front of us and ask them those questions. So that's one thing that occurs to me in addition to kind of get at the root of some of this stuff.

And we're creative, you know, heads of creative industries. I did, I was in LA just over the weekend and a lot of concern, as you, as you said, really great concern in the effect on Los Angeles already the number of jobs has been pretty substantially reduced and there's really concern of the impact here.

So which again, especially when, when some of these technologies are based on people's own creative output that then gets used. So lots to be on top of a couple questions. So as you said, many employers are already using these tools. What exactly Should. Exactly. But in your interview you touched on this.

But what should employers do before rolling out and then while using AI tools and hiring performance or really any kind of these tools you mentioned to make sure that they're being used in a fair way? I mean, you mentioned notification ahead of time. What other thoughts do you have?

I mean, ideally, if it were a union workplace, there would be notification, there would be a joint kind of development of what tools are needed, the procurement process, and then how they are used and deployed in the workforce. And there's many examples of that. WGA, IATSE, SAG-AFTRA and their bargaining have very specifically bargained over artificial intelligence.

Unite here for the hotel workers and hospitality workers has done a lot of great work over bargaining and technology, the communication workers as well. So there's a lot of examples.

The issue is that absent a union, the power imbalance in workplaces is so skewed to employers that it is very hard to have meaningful worker input, especially when it comes down to the bottom line.

And the tools that are out and marketed are generally ones that are going to, you know, increase profit, lower labor costs, increase productivity, often with little regard for the workers, because these tech companies, they need to sell the product and so that's what they are marketing.

And so I do think that's where the California State government, the Federal Government, has a role of stepping in and putting either some guardrails on the use of the technology or guardrails on the process for introducing the technology. Looking at basically, how do you kind of level the playing field a little bit given the power imbalance?

Do you think it's possible for us, for example, to get into these warehouses and look and see how these haptics or the things you say are being used?

I'm sure you could, I don't know how what the openness is, but you could probably get.

I mean, there's a number of Amazon warehouses, so I think probably you could tour a number of work sites and then there's a number of union work sites where you could also see how technology is being applied in a way that workers had a say in.

Okay, thank you. Just a couple more. Are there metrics you think policymakers should track to know whether AI adoption is helping or harming workers? Or along those lines, any metrics around job quality in AI forward workplaces or how that should be reported or.

That's a good question. I think obviously it's easier to track job loss, although the early numbers seem that there's not, other than in from tech companies. There's not mass layoffs but that there's more a lack of hiring. So attrition and that's going to have an impact on these young college grads.

And we're already seeing the unemployment numbers for young college grads inch higher. And so I think that's something that is worth watching. Is hiring slowing down as some of these entry level repetitive task jobs are being replaced. So that I think would be a big thing to look at.

Job quality is a lot harder to measure and takes I think more qualitative evaluation of how workers are feeling the impacts of algorithmic management. But as always, injury rates are one way you can look in some industries. And then these hearings are great.

And as much as we can get workers to participate or in district meetings so that you could hear testimony or have public comment from workers and the impacts on them.

Probably my last question. Do you have thoughts on how states can leverage the CSU, UC and community colleges to both create opportunities for both recent grads and mid career workers to prepare workers to use these new AI. Tools. You know and make sure that California kids lack of work get hired in some of these places?

Any thoughts on.

That's a great question. I know a number I can't speak to the CSU and UCs and community colleges although it'd be great if they start developing their own AI tools.

But a number of unions have apprenticeship programs and especially in more of the creative industries that includes will probably include the use of AI because it's become such a big part of what is happening in the industry. It's hard to say.

I mean one of the challenges of these AI products is that it's the only thing I know of where you buy a product on day one and it might be something different on day 365 because of machine learning. And so it is hard to say what will be helpful.

I think what needs to change institutionally is putting humans in charge of the technology rather than the other way around.

Say that last part again.

The challenge is having an institutional change or a change in our laws where workers are fundamentally charge of the technology and it's a tool rather than the technology being in charge of them.

Yeah.

And I think that will give young people a leg up because then they'll be able to go in and learn to use these tools to you know, be beneficial to be make their work better, faster and be more productive but not at a cost to themselves.

Yep. Yep. Well a lot of other thoughts but why don't we. We'll move on for now, but again, this is the beginning of the conversation. For sure. I appreciate you being here. Thank you very much. I'd now like to bring up Tara Fournier, the Deputy State Director of SHRM, the Society for Human Resource Management.

Thank you. Go ahead when you're ready.

Thank you. All right, well, good afternoon. Good morning Chair Becker, Senator McNerney, Committee Members, my name is Tara Fournier and I am the CEO, Founder and Chief Executive, HR consultant at T4HR Consulting and based in Southern California in Tustin.

I'm a 20 year HR professional and have been a Chief Human Resources Officer prior to founding my consulting firm. I am also proud to serve as the Deputy Director for the California State Council of the Society for Human Resource Management charm California.

Thank you so much for the opportunity to appear before you today to discuss the impact of AI in the workplace and how California can lead responsibly in shaping its future. As the voice of all things work, workers and the workplace, SHRM is the foremost expert and convener and thought leader on issues impacting today's evolving workplaces.

With over 340,000 Members in 180 countries and 26,252 Members in the State of California, SHRM impacts the lives of more than 362 million workers and families globally. SHRM sees AI not as a disruption, but as a defining opportunity to transform how work is done and how people thrive in it.

Used responsibly, AI can unleash new levels of productivity, creativity and collaboration across every workplace. SHRM welcomes the chance to help shape the policies that will make this future possible. Right now, AI is helping HR professionals work smarter with more efficiency, while still maintaining a personal touch throughout the onboarding process.

As one employer has told us, AI reaches out to passive candidates. For us, these passive candidates have diverse backgrounds and if not for AI, we would have missed them completely. As for the current State of AI in the workforce, it's no longer on the horizon. We know that it's already reshaping how Californians work.

SHRM's research shows that 43% of organizations now use AI for HR related activities, up from just 26% last year. These are not futuristic projections. They are real, measurable shifts that are happening in workplaces across the state.

In HR alone, AI is helping employers write more inclusive job descriptions, streamlining resume screening and improving time to fill rates for open roles. Nearly 9 in 10 HR professionals using AI report that it does save them time or increases efficiency and over a third say it does reduce hiring costs.

But most importantly, AI is helping organizations reach more diverse talent pools from veterans and older workers to those re entering the workforce, making hiring more inclusive. Another California employer shared with us we were struggling to fill entry level roles because of outdated degree requirements. Using AI to evaluate experience instead of credentials.

We increased our applicant pool, hired faster, and improved retention. This is what responsible AI adoption looks like in practice. AI enhancing human judgment, not replacing it.

As SHRM's President and Chief Executive Officer Johnny C. Taylor Jr. SHRM SCP said, Artificial Intelligence AI plus Human Intelligence HI equals Return on Investment ROI SHRM believes that while AI can help augment and automate routine tasks, we will reach our fullest potential by combining AI with human intelligence and oversight.

Through this combination of technology with human intelligence, HR professionals are achieving positive outcomes. SHRM's position is that AI can enhance efficiency, creativity, and fairness, but it should never replace human judgment. HR professionals are uniquely positioned to oversee responsible use of these tools and mitigate risks of inequity.

SHRM research finds that nearly nine in 1089% of HR professionals whose organizations use AI to support recruiting activities say it does save them time or increases their efficiency 36% say that using AI to support recruiting activities helps them reduce their recruitment, interviewing and or hiring costs and nearly 24%, one in four report that it has improved their ability to identify top candidates.

Organizations using AI to support learning and development L and D activities report that it has made their programs more effective, has reduced costs associated with these activities, and has increased employee engagement in these activities. But at the same time, most job seekers have not yet embraced AI.

Nearly 60% of US workers have never used AI to assist with job applications, and only 13% use AI frequently. This imbalance underscores the importance of transparency and fairness in hiring processes, which is why workforce readiness is very crucial. Responsible adoption of AI must go hand in hand with investment in people.

As AI reshapes job requirements, workers will need new skills to adapt and to thrive. SHRM and SHRM California support lifelong learning opportunities. These initiatives ensure that workers can continuously update their skills as technology and workplace demands evolve, ensuring that employers benefit from a more resilient workforce.

Employer Led Training Programs when businesses take the lead in developing talent, they can tailor their training to match industry needs or company specific goals, which fosters employee loyalty and reduces turnover while directly investing in the workers. Growth and public private partnerships that strengthen workforce readiness.

Collaboration between government, educational institutions, and employers create a more coordinated and efficient training pipeline. These partnerships help align curricula with labor market demands and expand access to quality training opportunities for diverse populations. SHRM recognizes that equitable access to artificial intelligence and related technologies is essential to building a fair and inclusive workplace.

As AI adoption accelerates, employers must be intentional in addressing potential disparities that affect underrepresented and economically disadvantaged communities. HR professionals play a critical role in ensuring that AI tools are designed, implemented and monitored with fairness, transparency and accountability. SHRM continues to advocate for policies and practices that promote responsible AI use while expanding opportunity for all workers.

Furthermore, SHRM earlier this year also launched their E Squared Initiative, which stands for Education to Employment Pipeline Initiative. E Squared was born out of the reality that while still the envy of the world, the talent pipelines here in the US from classrooms to careers are broken, leaky and busted.

Too many graduates lack the skills needed to succeed in today's workforce, while millions of capable people remain sidelined due to outdated systems, misaligned training and curriculum, as well as limited access to opportunities. SHRM's E Squared initiative is focused on fixing the fractured education to employment pipeline through awareness, advocacy and action. No single sector can solve this alone.

Employers, educators, policymakers and communities must collaborate to bridge the skills gap, expand access to career pathways and ensure workforce readiness. Public policy should especially support small and mid sized employers to keep pace with evolving talent demands. By investing in workforce development, California can ensure that employees remain relevant, engaged and capable of working alongside AI tools effectively.

In addition, it ensures that employers have a robust pool of talent to meet their needs and employees are able to support themselves and their loved ones. However, I would caution you and your colleagues about overregulation.

SHRM and SHRM California support a balanced regulatory approach that enables workforce, workforce and workplace innovation, safeguarding the rights of job candidates and employees while empowering businesses to make better workforce decisions.

We believe that a more targeted approach focusing on high risk automated decision making systems rather than basic tools that support human judgment and workforce management would better serve both employers and employees while advancing the goals of protecting consumers against potential algorithmic bias and discrimination. So our recommendations we have an opportunity.

California lawmakers such as yourselves have an opportunity to lead responsibly by providing transparency and accountability in AI tools with reasonable transition periods before penalties are applied. Clear standards will build trust among employers, employees and regulators by ensuring that AI systems are explainable and fair. Allowing sufficient time for compliance helps organizations adapt responsibly without stifling innovation.

Also supporting upskilling and reskilling programs for workers impacted by AI driven change. These investments empower employees to transition into new roles in industries rather than being displaced. A robust commitment to workforce development also strengthens economic competitiveness and reduces inequality. And finally, engaging HR professionals who are closest to these issues.

We are in shaping legislation that balances innovation with fairness. Our frontline experience offers critical insight into how AI is actually applied in Recruitment, retention, and workforce management involving HR leaders and policy making ensures that regulations are practical, effective, and responsive to real world challenges. In conclusion, California has always been the heart of innovation.

But leadership is not just about inventing new technologies. Rather, it is about making them work for everyone. HR professionals stand by ready to help. We see both the promise and the pressure that AI is bringing to the workplace. And we know that when people are prepared, technology becomes progress.

Let's make sure that as AI evolves, it does so with empathy, fairness, and inclusion at its core. Because when HI and AI work together, the result isn't just productivity, it's possibility. On behalf of SHRM and SHRM California, thank you for the opportunity to share our perspective.

We do look forward to partnering with you and your colleagues to build a future where AI empowers every worker in every workplace across California. Thank you. And I welcome your questions.

Thank you. Thank you. That was very informative. A couple quick thoughts and then I'll turn over to Senator McNerney. Yeah, it's interesting. I solved some of this when I was, you know, tried to start an online job company in 1994.

We were a little too early at the time, but, you know, the first wave of HR technology certainly was about making things more available, like job boards and things and, and making job notifications and all that easier to find. And, you know, I like what you said about enhancing creativity. You said, but combined with human judgment.

And so it'd be interesting to sort of kind of dive into a little bit more about how you sort of see that kind of playing out. I certainly liked what you said about experience instead of credentials. And I think that's an important shift in the. In sort of, you know, recruiting. Right.

And, you know, even, you know, here in our state government, I think we're trying to move away from it. You know, kind of the look for certain credentials has been a barrier to certain applicants.

And so if these tools can help us sort of look at the experience that people really have and bring, I'd be interested to hear your thoughts on that. And I like the idea of public private partnerships that enhance workforce readiness. So a number of interesting things to dive into.

But let me first turn it over to Senator McNerney, see what questions you have for our witness.

Thank you, Chairman. Thank you, Chair, for your testimony. There's a lot in here. I like a lot of what you said. Honestly, some of them are too wild about. But let's go over the positive stuff first. Responsible AI enhances human. Should enhance human productivity and not displace I think that's a great ideal.

And also, AI should never replace humans. It should enhance humans 100% on that. Absolutely. Now, what responsibility do you think companies should have that displace workers with automation? What responsibility they have to the people that are being displaced and to society at large?

That's a big question related to the society at large. I think that employers have a responsibility to provide their workers with as much training and information as possible as jobs evolve, as needs evolve, in order to preserve as much as possible. But I don't know that that answers your question specifically.

Can you clarify a little bit what you mean?

Well, you see this happening, companies. I don't want to point out Amazon, but I will plan to lay off a lot of people. And the CEO of Amazon is one of the richest people in the entire world. So what responsibility?

How can we sort of level the field a little bit so that we don't see such abuse of technology for the benefit of a very small number of people?

Yeah, that's a difficult question. I can't speak for Amazon, of course, but I would say that employers such as that would need to make sure they take care of a transition plan for those individuals. Whether that's giving them tools, resources, you know, certain, I don't know, a bridge to whatever they go to next.

I think that would be important for them to do. And I believe many companies do do that when they have, you know, separations.

Well, I mean, they're going to resist doing more than the minimum. And so I think it's our responsibility to try and protect those workers. And that's one of my biggest objectives here. How do we protect those people?

How do we even out that so that we don't see so much money flowing to a few individuals and a lot of people scraping by at the bottom as a result?

So I think education is really the most important thing in providing displaced workers, really all workers with the ability to learn how to use AI and how to change their skill sets to meet the skill sets of the jobs that are out there.

Well, I mean, that's an ideal answer, but we see a lot of these college graduates are facing a market where there aren't any jobs because the skills they just learned are now obsolete. So this has to be more than just education. There has to be some compensation, but not. Well, I'll leave it at that.

And that follows up. There's a clear difference between what research says is happening and is possible on the one hand, and what some of the CEOs are saying. So I'm going to give you some information here.

The New York Times reported internal Amazon documents that show the company plans to double sales by 2033 without adding any new US workers. So that's about 600,000 employees being replaced with automation or being prevented from being hired by automation. Amazon Robotics team wants to automate 75% operations, creating warehouses with very few people.

Now I've seen reports that there's factories in China with no people at all. The factories are dark because they don't. Robots don't need light. I mean we're heading in that direction. So we need to protect ourselves. Salesforce, the Salesforce CEO said that the company has cut 4,000 customer service jobs because AI can do all that work.

Elon Musk, one of our favorites, said yesterday that AI and robots will replace all jobs and work working will be optional. Now this is disturbing to me and how can we make sure that AI is not used in that way? And honestly, I agree with HR professionals are close to this.

They should have the best ideas, they should have the best informed way of developing these tools that don't result in those kinds of problems. But when we have CEOs that are using it in an abusive way, we have to find another way.

We have to have a stronger hand here in the state and federal offices to prevent that. So what is the tendency of companies to use AI in a not responsible fashion? So I like the idea of responsive use of AI. How many companies are using AI in a non responsible fashion? How do we characterize that?

Or do you have any information on that?

I don't have a specific number. I'll have to get back to you on how many numbers are using it irresponsibly. But when I say responsibly, I mean making sure that there's ethical use of this. And I think the way to answer a bit of your question, maybe this wraps up the previous question and this question together.

I think that there's things that companies need to do and that would be having an AI policy, ethical use policy, have human oversight. I don't know about the company in China that has no human beings. I don't think that's a great idea for us here in California.

I think AI plus HI is the ROI and you have to have human beings as part of these decision making processes. And I think having adaptability and cross functional skills training is really, really important in the company. Not just, you know, employers need to do those things. And I think that's, that's the prioritization that needs to happen.

Well, and I agree, but I don't think we can trust companies to do that without a little bit of a hand oversight to make sure that they are doing in some sort of guardrails, some sort of laws that make sure that most of the companies are using AI in a responsible fashion.

So my last question is kind of, you know, the Chairman, myself, all my colleagues, we're all elected. We have to answer the people that elect us. And there's a perception that AI is not really helping people.

I mean, there's a perception that AI is causing layoffs, that there's privacy concerns and that is going to drive decisions in Sacramento, it's going to drive decisions in Washington. So how do we move in a way that gives the public assurance that they're not going to be harmed by AI?

That's a very. That's a hard question to answer. If I had the perfect answer and you had the perfect answer, we would have done that.

But I definitely, again, I lean back on ensuring that we have a partnership between hr, between government, between public and private businesses to create a solution that does work for everybody, that there is collaboration and that there does need to be regulation. We agree on that.

Just not over regulation where it's too broad, that it harms things that don't intend for it to harm. Right. So things where it has decisions with people's employment, whether it's somebody departing the company or it's a hiring decision. I agree. We agree that that needs to be regulated.

So I do think partnership together to craft that responsible legislation is important. And we're here ready to help you with that.

Well, okay, I'm going to close by saying and give you a warning that you're going to be answering questions, asking questions soon, that there's we're both speaking the same language and the same ideals, but we're really kind of on different paths to get there. And that's not uncommon.

But we need to be more aware of what is the bottom line here. And that is making sure that AI doesn't harm our society and individuals that make up our society. So I'm going to have to leave that as a challenge. And I'll turn the microphone back over to the Chairman. Thank you, Mr. Chairman. Thank you.

Thank you.

I'll add a few more questions as well. Thank you. I mean, you caution against over regulation. Do you have examples of regulation that you feel goes in the right direction?

Yeah, that's a great question. I think I feel on the spot. Yeah. I will get back to you on which ones? I know SHRM is watching closely on a lot of them, and we do support.

Well, I would like to hear that. And I'm sure you. And do you track various other states as well? Are you California focused?

Well, I do. SHRM itself is SHRM for nationally. Yes, but I'm here for California.

Okay. You mentioned making sure human oversight. I mean, how should the Legislature be thinking about the degree of human oversight needed in AI tools and what are, in your experience, employers currently doing to ensure the tools do have active human oversight?

Yeah. So human oversight would be, you know, anything that at a decision point with someone's employment would be an area where you would want to have a human being.

So anytime somebody is being hired, anytime we're assessing risk, if there's a decision to be made in a business that's a critical decision that impacts workers, there should be a human being in that place, in that spot. And so having that level of oversight works.

Things like payroll systems and other things that are automated, maybe it's a different story. I think we have to look at each of the different types of tools that are being used and see whether or not there's a major decision point being made there and look at it.

But I think sometimes when we're too broad about writing the legislation, it might inadvertently cause challenges in other places where it doesn't necessarily need to to do so.

Well, I certainly believe that most employers are using AI and hiring responsibly and do have human oversight. But I do challenge you to help us think about what to do about the ones who might not be doing that. And that was from the subject of this legislation that was attempted last year.

We have to figure out how do we address those who are not using it responsibly or in this case, don't have human oversight of those kinds of processes. So that's sort of a challenge. I'll leave this sort of challenge you to help us think about that a lot. I could ask about skills.

Let's see if you have any thoughts on what training or organizational changes make gains in productivity more from AI, more likely. And do you have any thoughts how the state can help facilitate those gains?

Yeah. So I think the train, the training that I'm. What I'm seeing personally in my practice is when we have employees go through and learning how to effectively use a prompt, how to effectively understand the way that AI works, we're getting much better results.

A lot of my clients now are putting AI ethical use policies in place in their employee handbooks and Regulating in that, in that regard, but having support from the government and through legislation to help maybe put some guardrails around those that maybe aren't as sophisticated quite yet may be helpful just to shape and kind of push towards that direction.

Okay, yeah. Maybe two last ones. You mentioned the E Squared initiative, so maybe you can tell us a little bit more about that and what specific policy levers would help strengthen the pipeline between education and employment from a employer's perspective.

Yeah, thanks for that. So E2 or E squared is our initiative. And what we're trying to do is help people in all different places, whether they're displaced workers, whether they're second chance coming out from, you know, previously incarcerated under even older workers that haven't had an opportunity.

It's about providing the tools and resources and training that they need to understand how AI is impacting and the work that they will need to do when they come into the workplace again. But for those that are already in the workplace, SHRM has certifications that can be taken to learn how AI and HI can be together.

So there's a lot of work to be done and a lot of work and tools that we already have in place at SHRM that we're happy to share more with you, some materials after this hearing.

But definitely that's what we believe is having work skills and the education on how to use the tools and, and how to use automated artificial intelligence is the way to go.

Yeah, yeah. Recommendations for soon to be college graduates. Hopefully college graduates. Let me know.

I've got a couple too.

I guess. Lastly, given that over half of Californians do hold an unfavorable view of AI, what is your view on how, if it's possible or desirable, I think from your point of view, how would we build that public trust and AI at work?

Building trust, I mean, building trust is the hardest thing to do, right? Trust is currency. And I think being transparent and being able to allow multiple viewpoints to come in, just like my colleague Sarah talked about before, is bringing people together to learn and discuss and have a discussion about how it works.

I think people are afraid of AI and that it will displace and things, and I recognize that it will. But it also is an opportunity for people to evolve and to learn new skills and to figure out ways to work.

And so I think people, the way to build trust is to not have it be the big elephant in the room, but allow everybody to listen and to hear and to input so that they understand where we're going and what we're trying to do and the intention behind why we would be regulating and having government assistance in creating those guardrails.

Okay. Yeah, helpful. I think one, you know, one thing going forward that, that we can do again from our, you know, of you here Senators and here in the Senate is to, you know, is to bring folks together as these are rapidly evolving and to discuss.

One of the bills I done two years ago was a K12 AI and K12 education Bill. With this, the Superintendent, Tony Thurman, we did that as a work group Bill and we had the first workgroup about a couple months ago and it was really fascinating.

Just hear teachers from across the state talk what they are doing to integrate AI in their classroom and how they're using it and to kick off that process of how it can be used. And if they don't want to use it in the classroom, what they're doing to prevent that too, as well.

But really learn from each other and work together, encountering this brave new world. Thank you for being here. Thank you. And look forward to being in touch.

Thank you very much.

All right, thank you all for your patience. And we'll now have our final presenter, Molly Kinder. Molly is a senior fellow at the Brookings Institution. And is Molly going to be.

Can you hear me?

Yes.

Oh, great.

Okay. There you are.

Great. Well, I'm so sorry I can't be there in person, but I appreciate you tolerating me being virtual and joining this fascinating discussion. To Senator McNerney and Senator Becker, thank you for putting a focus on this incredibly important issue and for including me in the conversation.

I am a senior fellow in Washington, D.C. at the Brookings Institution, and I swear I have the greatest job in the world because every day my focus is in researching and studying how gen is impacting work and workers. And I love the second part of my job and what do we do about it?

So I'm going to share a little bit of my recent research, some of the data, the findings about where we are today, and then a few ideas just to spark some discussion of when I think about California specifically, what are some priorities and what are some big ideas.

So first, just overall, my philosophy on AI and work is the reason why I do this job is I think, for workers, AI pretty much presents both wonder and worry. And we've heard a mix of these two things in the previous discussions.

I talk to workers all the time as part of my job across the country in a range of different occupations. I talk to industry, I talk to economists. I talk to labor unions. I was at the Vatican last week. And what I hear is a mix of concern both about this idea.

I think a lot of folks who've never really worried about technology being something that might. Might be a threat to their livelihood. I think there's this little fear inside a lot of people's heads. If this technology gets better and better, could it potentially do my job? Is my livelihood at risk?

There's a lot of anxiety about young people. I'm going to talk about that more in terms of the data and the potential Solutions. I have three young kids, 10 and under. All the time I'm hearing from parents a sense of being befuddled. What in the world do we do to prepare our kids for this future?

Our recipe for success of making sure your kids study and go to college and get a good job all feels really at risk. But there is a lot of opportunity. I use AI all the time in my job to make me think deeper, to brainstorm about the hardest issues, to edit my work.

I think AI can be a tool that really is something that workers could truly benefit from. So I am neither all doom or all gloom, but I'm clear eyed that there are both risks and opportunities.

What I want to first talk about is some new data that some colleagues at the Yale Budget Lab and I just put out.

Just a few weeks ago, I think three weeks ago, we published a paper and what we wanted to do was to take all the headlines and all the buzz and all the pontification and just ground it in reality.

And we asked what has happened to the labor market in the just about to be three years since ChatGPT was launched at the end of 2022. And what we did was we combined the best government data on the lab with the data that OpenAI shared with some of their exposure.

So the extent to which AI could be used in different jobs. We also looked at anthropics data on usage and we did a ton of different sophisticated analysis. And basically we wanted to know is, are we in a jobs apocalypse? If you zoom out and look overall at the labor market, are we seeing signs of disruption?

Are we seeing signs that there are movement from the most exposed jobs, for instance, to less exposed jobs? Because we would imagine if AI was causing mass disruption, we would see people leaving those higher exposed jobs and going to safer jobs.

We might expect to see in the unemployment data that some of the folks who are unemployed come from the more highly exposed occupations. And basically we found none of that. We found a story of remarkable continuity with the caveat as what I'm describing is really zooming out and the broad labor market.

Our methodology would be able to tell if the whole house was on fire, not if there was sort of a fire that broke out in one room on a stove. It's entirely plausible and likely that there may be some specific occupations, some specific individuals who have felt the impact of AI in terms of job losses.

But when we zoom out to the full labor market, it's not enough to truly be discernible.

And so my headline for this is borrowing from the wicked movement movie that so far AI has not been able to defy the kind of gravity that most technologies, even that end up being wildly disruptive General purpose technologies, they still take a while to fully diffuse in the economy and to start showing up in labor market impacts.

We compared the pace of disruption from Gen AI over the last three nearly three years to the pace of disruption in previous waves with computers and the Internet. Gen AI is a little bit faster, a little bit more disruptive, but mildly so. And we looked.

It's not clear that that trend really started with ChatGPT it could have predated it so far. It has the same gap between how quickly the technology is advancing and frankly a little bit of the friction in the workplace that's resulting in very lumpy adoption. You've got.

We contrast exposure to adoption and show that there is wildly uneven adoption across different sectors. Some sectors moving extremely fast and adopting AI and potentially, you know, seeing real job impacts and others that in theory could use the technology but are moving much more slowly.

One really important exception is we did look at this was a conversation that I know the Senators have have picked up already on early career and our data is consistent with the Stanford paper. This canaries in the coal mine paper. We did find disruption amongst young workers. It's not clear from our data that that's because of AI.

So we know there's something happening with young workers. Our paper did not prove that AI is the culprit. However, when you put together a lot of these different analysis, a lot of my discussions out in the real world, I think there's enough for us to start really taking this very seriously and starting to think about solutions.

And I'm happy to talk a little bit about that very quickly. When I think about California, a few observations about why this issue of AI and jobs is so relevant in your state and also a few things where I can see California really leading on. The first is from looking at the data.

I have this terrific colleague at Brookings named Mark Muro. He's a senior fellow on our team. He and his colleague Sriya and myself, we did some analysis looking at the geographic footprint of AI exposure.

It is very clear that California is going to experience AI's impacts probably faster and probably more profoundly than most states across this country and particularly in your metro area around Silicon Valley, off the charts exposure. And when I think about some of California's industry, a lot of Cs come up to me.

Computers, creatives, cognitive, to some extent also clerical. Over 10% of your workforce is in clerical and customer service roles. These are extremely exposed sectors. And to the extent AI is going to be beneficial to workers, California workers are going to feel those benefits more deeply than a lot of states that don't have as much exposure.

But to the extent the path we're going on is potentially harmful to workers and will lead to displacement, California is really going to feel those effects. California is a leader in the country in legislating around AI. I'm so impressed with some of the regulations, the safety, the privacy, all the various things you're doing.

I'm really hopeful that California can also be a leader in some of the solutions specifically around workers.

And I think that's probably going to be some combination of regulations, but also probably piloting some new solutions and happy to talk more about what some of those ideas California is not only likely to feel the impacts in terms of its workers from AI, but your state is inventing the future of AI and it's going to impact workers everywhere around the world.

And I firmly believe that there's opportunity with AI to be good for workers. I think there's lots of ways AI could be designed to really enhance our capabilities. I love, for instance, open evidence as an example of it's really meant to help doctors not do their job, but improve their diagnostic reasoning.

Most of what I hear in all the podcasts I listen to coming out of VCs and the Silicon Valley industry are not that it's very much a replacement lens. I don't think that's necessary. It is definitely where the incentives lie. I was at the Vatican last week.

I had the honor of participating in an amazing workshop at the Vatican all around AI and something the Vatican's calling a new digital rerum navarum.

This idea that this new Pope Leo is likely to update the last Pope Leo's groundbreaking encyclical that he put out during the second Industrial Revolution, really calling out ways in which workers really paved the way for lots of safeguards and workers rights during the industrial Industrial Revolution. Big question, what would that look like for an AI era?

And one of the things I said was my hope that Silicon Valley can do more to make sure that the kind of innovations they're putting out for workers are beneficial to workers. And something that I think is getting really overlooked is a call from Pope Francis to make sure AI lifts up the least among us.

And what I said in my remarks last week is that that's not the way I see AI playing out right now. If you look at usage and anthropic had some great data AI, AI is primarily being used by the most privileged, the most educated knowledge workers.

They're reaping the benefits, not those with the least opportunity and the least education. I think there's some really exciting ways that that same spirit of innovation that's happening in Silicon Valley can be harnessed to reimagine jobs and opportunity for lower wage workers. But it's not going to happen without some real intentionality.

And California can be a hotbed for steering AI in those positive Just 23 more quick points because I'd Rather go to discussion. The other inspiration I take from California is the very first case study I wrote that was worker centered was on the Hollywood writers.

I wrote a long case study for Brookings, spent some time in La. I think California, through the writers have proven that worker voice and worker power is just extraordinary powerful in making sure AI truly benefits workers and avoids harms. And their pLaybook is phenomenal.

Of course, the challenge is very few workers are in a union that looks like the Writers Guild. And some research I put out at Brookings with colleagues calls this the great mismatch.

The kind of sectors that are most exposed, most likely to be disrupted by AI have the lowest union density in the entire economy, often as low as 4%, 1%.

What can California do in a bold way to take that inspiration from the Writers Guild, but figure out about inspiring, innovative, outside the box ways to get more power and voice for workers at scale in these places where we simply just don't have that union density. This is a place where I think California can really innovate.

The rest of the country struggling. We don't have great models. I'm inspired by the fast food, the fast food work you've done with the worker council. Is there something like that for AI that California might get to innovate? I think other states would quickly adapt. Two more quick things. California sits.

The biggest AI companies in the world sit in California. Researchers like me, in order to really know what's happening to workers, need their data. Anthropic is the only company that's really transparently shared all their usage data, including their enterprise data. We desperately need that from the others.

Is there a role for legislators to help push prod, really get us that kind of data that I think could be really transformative. And last, I just was so inspired by the conversation about employers years. What does it look like for employers to take the high road? This is not just a story about reskilling and retraining.

You know, the public hear stories of how much CEOs are saying they're going to cut workers. The Morgan Stanley research put out a an analysis that said if AI was fully implemented across the economy, the S&P 500 would save close to $1.0 trillion a year, primarily through labor savings.

I don't think that sits well with the public public. What should our expectations be of employers specifically on this early career question? I'm working. I've been working for a while on an essay for the New York Times that I hope is coming out soon with a big idea for carrots and sticks.

How can we really push prod, maybe even force employers who are shedding these entry level workers to make sure they're still training in the next generation.

I think we need big bold ideas like this that push, push employers, push the incentive, push the tech developers in a direction that is much more pro worker than what we're seeing right now. Thank you very much.

Thank you. Thank you very much. I appreciate a number of things. You said the transparency of data, that's something we can pick up on kind of citing the Hollywood workers and that's interesting. So sort of kind of had this sense intuitively what you said about the kinds of industries most exposed have low union density.

But to see it reflected in the data is really important. You know what you said and I don't know open evidence, but you know, the examples where AI really can amplify our important to note and to understand. You know, I ran a technology.

We kind of really created the category of legal analytics and that was all information that was not available before about our legal system, but using really more natural language processing than anything else. But you know, AI is such a broad term.

We were able to mine through millions of pages of legal documents to extract that information about the judicial system that people nobody knew before and provide them that to lawyers to use. So anyway, just there are interesting things there, I think to talk about the industrial Revolution.

I read a book recently about that time period and initially it was, it did result in child labor and workers being devalued and such. And as you said, ultimately policy came through and it's interesting to hear about your trip to the Vatican on that line.

So anyway, number Power in Progress by Daron Acemoglu. Is that the book you read?

No, this was, I think I do have that one to read. This one was particularly about England and you know, when they talk about 20% of the workforce was in the making of cotton and, and so the implications of technology use in that time.

So anyway, so a number of questions, but first again I'll turn it over to Senator McNerney to lead off.

I think the Chairman. Thank you, Molly. Very interesting presentation and I want to a couple of questions. Your research shows that major job losses haven't happened yet. But what are the warning signs that we should be looking for if that would start?

I mean, how can we look to the future based on existing data today and tomorrow to see if we're facing a major crisis with displacements?

That's a great question. First, I would say my colleagues at the Yale Budget Lab who I teamed up with the paper plan to every single update the data with the latest labor market data to be that sort of monitor.

I will say our data is imperfect because when we have to rely on these theoretical exposure metrics, there's a lot of exposed jobs that really aren't yet using AI. So it's too broad, it's not sensitive enough.

Part of the reason why we're really calling on the AI companies to share with us their usage data is that's more forensic and it would allow us to better measure what's really happening. You'd want to isolate what you're measuring on who's really using AI to figure out is that causing any labor market implications.

But my colleagues and I are very dedicated to continue to measure this as transparently and as effectively as we can.

Just a few weeks ago I was in San Francisco with a really interesting all day scenario planning workshop app that was a lot of preeminent economists who were in town for an nber, a seminar on the economics of transformative AI. So Eric Brynjolfsson from Stanford was one of the organizers.

And actually that was one of the things that we were tasked to do was to come up with the metrics that would allow us to know what world we're in. Are we yet in a world where AI is likely to be disruptive? I think it's my headline here is it's still very early days.

We probably shouldn't expect to see really transformative results yet. I wouldn't be doing my job if I didn't think those are coming. That's why I take this very seriously. But it's still, I think the headline is it's still very early days. What I'm doing is I'm tracking the kind of canaries in the coal mine.

Where do we think we're likely to see the impacts the first and the fastest. And that for me is early career is something I've been writing about for a year. I had an essay in Bloomberg a year ago warning about why I was concerned AI could kill the career ladder.

I think that we're really closely tracking this because I think it is could very well be one of the first things that sort of collateral damage from AI. And if this is really the case, we need some good ideas.

So I have a report I'm writing for Brookings right now that's going to come out very soon with a whole suite of ideas for how we might start addressing that.

I think the other thing is to keep really close eye on some of the jobs that are likely to be some of the earliest movers of, of impact, customer service reps is one that I'm very closely monitoring. The technology already is good enough to displace a lot of those jobs.

Creative industries, writing, copywriting, marketing, computer science, of course, is something everyone's tracking closely. So, you know, keeping our ear to the ground on where the usage is going the fastest, where we're likely to see the biggest impacts, and making sure we're measuring those as best we can, I think is probably the best way we can do that.

Well, thank you. And you are sort of having a very optimistic outlook and I appreciate that. And some people, some businesses, organizations say that AI will create new jobs and raise wages. But based on past trends, will that happen without government intervention, without some significant guidance from Sacramento, Washington and other state capitals?

I don't think that companies will do this on their own. But please disabuse me if I'm wrong.

I hope I'm not sounding too optimistic because I think what I'm trying to do is my data is meant to say that some of the hyperbole we're hearing that, that, you know, tomorrow we're going to have drop in workers and we're already in AGI that's not yet appearing in the labor market data.

And I think that should give us some reassurance and think as policymakers, you have a little bit of time to act. But I think this technology is extremely transformative. The potential of this, and I don't necessarily think the path we're on is one that looks like it's necessarily going to be good for workers.

So I think the future doesn't have to look like the sort of more muted labor market impacts of today.

I think this is a really great question and it's something that I've been asking a lot to fellow, you know, economists and experts is, you know, we have a sense of if this technology gets good, the kinds of things that might shrink and go away and if we've learned anything from deindustrialization, is that even if you have macro job growth, it's not at all clear that the job that emerge are the preference, the possibility, or in the place of the people who've lost their jobs.

And I think what worries me about this technology is it's a cognitive technology that for the first time is really putting at risk very good jobs, jobs people hold dear. They've invested a lot of money in education. We're not yet seeing mass displacement of these jobs. But there's the potential in there.

I just was recording a podcast interview before I joined with some economists who just published a paper at Berking who are predicting that unlike the past hundred years, technology may no longer result in higher paid, better jobs. We might be moving into a world where the jobs that grow are less quality, lower paid.

And so I think there may be a point where we have to ask questions of what's the good jobs agenda? What is a role, what's a more muscular role for policy in making sure that the jobs of the future are actually good jobs? Everyone thinks the answer is just reskilling.

Well, what if the jobs are reskilling for are not good? So I think there are some big questions. I think part of what we saw in the industrial strategy of this last presidential Administration, I think was a little bit of a job strategy for the people who lost jobs in deindustrialization.

What would that look like for a cognitive industrial revolution? What kinds of opportunities do we need? Do we invest more in a country in the social sectors? Do we invest more in small businesses? Do we. This is an idea that I'm putting in the paper I'm writing.

If we know that the creative industry and maybe even journalism could be the collateral damage of AI's business model, maybe we need to be taxing the proceeds of that and really investing in the creative industry and in journalism. So I think these questions of, you know, what is the.

What kind of jobs do we want to make sure arise? What would be some industrial strategies, economic development strategies, some big bets from government that might help us get there? I think is exactly the right question to ask.

Well, we look for you, look to you to help us find some of those answers then. I appreciate that. So one of the things you said was that adopting AI hasn't resulted in big disruptions yet.

But what I'm wondering is, are you talking about AI alone or AI combined with robotics replacing factory jobs, or AI combined with new automobile designs to displace pay drivers or teamsters? I mean, combining AI with technology with other technologies is really where the disruptions are going to be coming from, in my opinion.

So could you see, could you answer if that's already happening in your data? Are you looking at that, that or is that something we need to think more about?

So the first thing I would say is our data says nothing about the future. It's not a prediction. I have other reports I've written for Brookings looking at predictions. So what are the jobs in the sectors that are most exposed to gen AI? That is very much forward looking.

This paper was meant to give a snapshot of today to answer basically to disprove the null hypothesis that the labor market is already in the midst of a technology driven jobs apocalypse.

And the way that we measured that was to look at are we seeing churn in the labor market, Are we seeing people moving more than they normally would into different types of jobs that correlate to AI exposure?

And we look at that from the moment ChatGPT launched to today, and we compare that level of churn and disruption in the job, you know, who's employed. And also we looked at who's unemployed to see if there's some correlation between those employed and AI exposure.

And our headline is we're not yet seeing that at scale, which doesn't mean there aren't some customer service reps who have lost their jobs, or somebody in electric vehicles, or a copywriter or young people. We looked really broadly and if the labor market was experiencing really significant discernible disruption, our data would have picked that up.

There's other things that are more forensic. You know, if you're just looking at, at the EV industry, you know, our data is not going to just be able to sort of just pick out that we were trying to look at. Are these headlines that make everyone so anxious to think, you know, we're hemorrhaging jobs today.

Do we find that in the data? And the answer is no. A separate question is moving forward, are there risks? Are there hazards? And I would say absolutely yes. I mean, it's not just gen AI in cognitive roles and creative. Certainly robotics would accelerate job displacement in a whole different set of roles.

And that's why we're going to continue to track it. But really the purpose of our paper was to do a gut check of where are we today and does the sort of anxiety. There's a lot of anxiety in the public about the future and I think that's probably fairly grounded in true concern.

The question is, are we already seeing this in the data in terms of the labor market? And the answer is no.

Okay, thank you. One other thing, my last question is you mentioned that you're getting pretty transparent data from anthropic, but not from much of anybody else. How would we compel, how would we in the government compel companies to be more transparent with their data? I mean, AI is incredibly difficult to understand.

I mean, there's no way that we can really understand the decision making process in AI. So how would we go about compelling if we were to do that? Say OpenAI or Microsoft or Meta to be more transparent. What would that look like?

Well, first I'd say that that's a little bit beyond my expertise. So I would actually, I would. I think you're probably more an expert of that than me. My, actually, my hope is that the companies are going to voluntarily do this. We're having some really productive conversations.

So my hope is we wouldn't need to compel if it becomes a real problem and we're not getting enough transparency. There is a question of whether or not there is some legislative response. Again, it's a little bit out of my expertise of how you would actually structure that and do that and whether it's meaningful.

The piece of data that researchers really want is similar to how Anthropic has published and made it openly available. Their data set, some of the usage data that's been extraordinarily useful to researchers like myself as well as. And they also included their enterprise data. I know these things are a little bit hard for companies to do.

It's not just a lack of willingness, but my hope is they're going to voluntarily do it on its own. I think there's a broader question about are there other pieces of data that would really benefit policymakers and the public to know about how AI is impacting the labor force?

I don't know that this is the right idea, but for instance, companies, companies don't report their early career hires in their SEC filings. They have to report their overall headcount. But as a researcher, I don't have any way to look at the major employers and know from the outside whether they're shedding workers.

You know, I, I don't know, you know, some of the investments they're making. So, you know, again, maybe this isn't the right tool, but there could be a conversation about what sort of data do we need to either hold employers accountable for how they are deploying AI vis a vis their workers.

What kind of data from the AI companies would help us better understand what's happening?

Well, I'm asking you that question. At any rate, I'm going to turn this back over to the Chairman. Thank you, Mr. Chairman. Thank you.

Well, yeah, that can be part of our ongoing follow up. And as you said, you are going to be reporting monthly on the impacts on the labor market. So we'll look forward to. Please share that with me. I will definitely share that with our. Committee and Senator McNerney and others.

The Senate caucus does have a consultant as well. Make sure we loop in as well. Yeah, you mentioned a couple things I'll comment on. It's interesting to mention sort of the infrastructure Bill. Right. That was an attempt, attempt to say, hey, you know what, let's build things here and create those kind of jobs here in America.

And it's interesting you sort of saying that. Like what, you know, kind of this good jobs agenda that you mentioned is something that we'll certainly keep following up on.

I did want to mention since we called out one company, specifically Amazon, around warehouses, that they did make a 2.1.0 billion commitment, they say to educate and skills training for 50 million people by 2030 for the future of work. So we will keep an eye on that and you know, invite them as well.

You know, again, there's been ongoing dialogue with companies about what things that they are doing and things that we maybe think they should be doing going forward. So that was valuable. I wanted to note that. And also invitation to tourism of the warehouses, which is something that came up earlier. A couple last questions for you.

I mean you mentioned sectors that was helpful and I noted that you said customer service, some of the creative industries. Any other mention? I mean we've seen translators, entry level coders.

I do feel in districts I represent, which again is the heart of a lot of technology, that there's a lot of conversation among parents about where they're, you know, what careers to prepare their kids for. Right. And a lot more focus also on physical careers. Any other thought of other industries we should be watching?

Yeah, yeah. I mean the one that. It's a little bit of a personal passion of mine I'm writing. I haven't yet published this report, but I've been working on it for a long time. High school educated women in clerical roles. So customer service is one.

But think HR assistants and legal secretaries and medical coders and you know, folks working in back offices. We have a ton of data to show these are the best jobs for high school educated women. Best pay, best mobility. You can retire into them, they're gentle on your body.

And I worry that Gen AI could be to them what deindustrialization was to high school educated men. There's really only a story about automation there. It's moving a little bit slower because in order for companies and employers to really automate those roles, they have to really embed this into workflows.

It takes a little while, but this is a really big deal. And as a country we're very focused on the loss of male manufacturing jobs, as we should be. But the same thing could be happening to women and we do not have a plan for that.

The jobs that are rising and have been rising are very low paid jobs like home health aides and fast food jobs. And I think this redoubles the urgency of making those better jobs.

I'm working on a pilot to figure out can you harness AI to create almost like a home health aid plus role, a more clinically, you know, advanced job that could be better paid? Is that one way we could be thinking about improving those jobs?

But this is definitely something I just want to put on your radar because it's fallen out of both political and philanthropic topic focus. I don't think most people know about that early career. We've already talked about, you already know creative and obviously, you know, coding jobs are under pressure.

I think in some of the private sector industries that are moving the fastest on deploying AI in ways that I think could, I wouldn't be surprised to see impacts on workforce or finance. For sure, marketing and, and sales is a little bit lagging.

But I would say in some of these, you know, finance and marketing are highly exposed and I think you're, I would not be surprised to start seeing job impacts.

Yeah, well, as you say, well, you know, sales, you can imagine some of the entry level, you know, some of the calling jobs which are level jobs in that world. Right.

To your point about the career ladder and to your other point, that's why we've worked hard in California to raise the pay of fast food and some of those other lower skilled jobs and as you say, make them better jobs. Let me see if I have the last question here.

There's sort of a question about the rate of occupational change being compared to other technological changes. And you know, but the level of investment going to these companies is just massive compared to other and much faster compared to other technologies. Any quick comments on that?

Yeah, no, I, you know, there are some ways in which it feels like gen AI should move faster. You know, it's, you don't need to buy a big new expensive computer, it can diffuse faster. I think part of the sort of reality check from our paper is how uneven the adoption has been.

So there are some industries where there's very limited, there's little friction. It's easy for a coder to use cursor, it's just easy. As a researcher, I can immediately use these tools. There's nothing, it's no proprietary data, you know, I can just immediately jump in and use it. But there's a lot of sectors that are more highly regulated.

There's more resistance. So it's just been very uneven. And I think some of our data might be lending a little bit of support to this framing from some of these Princeton academics. AI is normal technology, meaning it is likely to be very disruptive to the labor force. But we do not have drop in remote workers already.

It's not totally defying the gravity. That diffusion just takes time. The way I've been framing it is I expect a lot of lightning strikes. I think that AI is going to move really fast in certain sectors. It's going to have a very big impact on certain types of workers and you're going to see much longer lags elsewhere.

I think it's up to us to be tracking that and being really prepared. Again, the data is reassuring that we're not yet in a jobs apocalypse. But I wouldn't be doing this job every day if I didn't see enormous potential for disruption.

And I think to some of the questions that were so smart in the last session, the incentives of employers right now is to cut costs. It is very much a focus on labor savings and they're under extraordinary pressure from their investors.

And while there is potential to augment and maybe the fact that we're not seeing huge impacts in the data is because maybe these early days are a little bit more augmenting. I think it's very easy to see a future where employers are cutting costs and really we see a labor market impact.

And I think really the questions we need to be asking is what, what are the levers we can pull that steers this in a way that's better for Workers. And I would just emphasize if we hang our hat on reskilling, we're in big trouble.

Yeah, well, lots to think about there. One thing that just comes to mind against my background too, is just sort of we think about small business. We think about also entrepreneurship. Right. These technologies also lower the barrier for entrepreneurship.

And to extend that can be a piece of the puzzle of really encouraging, making it easier, helping people start their own businesses as well is something to think about.

Anyway, I just want to plus one that for sure. I mean, my dad was an entrepreneur. He started a small business when the Internet happened. Was it sort of one man show? Just small. I think one of the questions I have with some of the impacts on early career, is there something related to that?

It's never been easier to start your own thing, but who's got the resources for that? And sort of what can we do to make sure more people can seize that opportunity? So I just want to plus one that.

Yeah, good. I was just talking to my friend, Dilawar Syed. He was from our area and then went to. He was running the small business Department under, under Biden and now is teaching and, you know, again, one of those kind of people that we'll be looking to talk with going forward in this effort. Thank you.

Thank you for joining us. Really fascinating stuff. Really appreciate it. And with that, we'd like to move to public comment. Appreciate those who are with us here and we'll give you a little bit extra time since we have it. We appreciate it.

Good afternoon, Chair. Let me take my glasses off. Chairperson Becker and Senator McNerney. My name is Kathryn Kennedy. I'm a registered nurse. I've been a nurse for 45 years and a President of California Nurses Association, the largest nurses union in the State of California. Nurses have been confronting the impact of algorithmic technologies, including AI, for decades.

And we know that without exercising our collective voice as workers, employers will use technology to descale nursing, speed up our work, and ultimately replace us when it comes to healthcare without the guardrails, Employers will use AI and other technologies that will interfere with our ability as nurses to exercise our critical thinking skills and our professional judgment to do our job providing safe patient care.

Employers have inserted AI throughout our hospitals and the electronic health records, clinical diagnosis algorithms, remote patient monitoring system, excuse me, predictive staffing tools, to name a few.

Hospitals deploy AI tools to determine when patients should be sent home, to answer patient emails, shift change reports, and to conduct emergency Department triage, where an algorithm decides when a patient gets seen by a Doctor. Instead of a nurse assessing patients at regular intervals. And this is just the tip of the iceberg.

When AI tools make decisions about patient care, these decisions must be viewed as decisions about a nurse's working condition. Hospitals use AI staffing algorithms to determine how sick patients are and the level of staffing that is needed. However, these tools often underestimate the staffing needs. Assigning too few nurses, making our jobs that much harder.

AI has not resulted in the efficiencies and the time savings for workers that the developers and the employers had promised. What we're seeing is that nurses and doctors are taking more time, not less, reviewing AI notes, correcting errors and hallucinations made by these technologies.

AI hallucinations in healthcare work means delay and or missed care medical errors, algorithmic biases that harm patients and can cause lives. It also creates a huge liability risk for workers. Who is liable for the harm that might result from a new technology? The most convenient scapegoats are going to be the workers who interact with this technology.

That includes nurses and other frontline workers. And then it will be the human reviewers who are then sent in to create a facade of safety. Therefore, employers and technology developers must be held accountable for the harm that AI may cause, not nurses who would be disciplined by management if they raise such concerns.

AI can never replace human judgment and compassion that defines nursing. Protecting the healthcare workforce means protecting nurses clinical judgment and our ability to act on that judgment without being overruled or undermined by an algorithm.

So thank you, thank you.

Thank you and good afternoon, Chair and Senator. My name is Diane McClure and I'm a board Member of the California Nurses Association, and I've been a registered nurse for 36 years. Our union has been proactive over the years in calling out risks AI and other new technologies can pose in our work as nurses.

From algorithmic biases and other adverse patient outcomes to the risk of deskilling and replacing nurses professional judgment. In addition, employers are using AI surveillance technology to monitor both patients and nurses, which further erodes patient care and workplace conditions. AI tools record our conversations, capture patient data, and send that information back to our employers and outside vendors.

This surveillance allows employers to track how long nurses spend with patients, whether we are moving fast enough or even justify cutting staff. Hospitals are also experimenting with ambient monitoring systems that use AI along with a slew of cameras, microphones and other sensors to track what both patients and nurses are doing.

They can monitor and record facial expressions, air, temperature, movements in a room, and speech. Some hospital employers are even requiring nurses to download AI ambient listening apps onto their personal mobile devices. This AI data collection raises serious privacy concerns for both healthcare workers and our patients.

But increased worker surveillance leads to distrust and undermines cohesion within a unit and also threatens union organizing because people are fearful of what they say in conversations. This surveillance also impacts patients as many big tech companies partner with both hospitals and governments to develop and share surveillance and data collection tech.

For example, Palantir is partnering with corporate hospital chains to maximize profits by developing AI software to automate staffing, patient care plans and billing. Palantir is also working with the Trump Administration and ICE to compile data on immigrants for their deportation machine and to make it easier for federal agencies to surveil all of us.

Big tech surveillance in our hospitals also erodes the trust between nurses and our patients. Patients avoid communicating openly with nurses or avoid care altogether, compromising our ability to provide safe therapeutic patient care, making our jobs even harder.

As you investigate AI's impact on workplaces, nurses urge you to safeguard our workplace rights and patients rights to high quality care. We must ensure that hospitals and other places of healing are not turned into data centers for corporate profit.

Thank you. Thank you very much.

Thank you. Chair Senator Jp Hanna, Legislative Advocate on behalf of the California Nurses Association. Going to echo the comments of the two nurses we had here today. As bedside nurses, CNA Members witness the harm to workers and patients that can arise from algorithmic decision making in the healthcare sector.

When healthcare employers deploy AI and other algorithmic technologies, what employers call efficiency often means doing more work with fewer people. Now in the healthcare sector, tech companies are selling AI as the next great tool for time savings, cutting labor costs and increased billing.

But the reality is that nurses are actually spending even more time reviewing, correcting and fact checking what this technology gets wrong. Hospitals claim that AI will relieve workloads, but in practice, it adds new tasks and new risks.

A study from the University of San Diego last year found that Clinicians using a large language model to reply to patient messages actually spent 22% more time checking messages for errors. When an algorithm makes an error, it is the nurse who is blamed, not the software.

When a chatbot fabricates information, it is the nurse who is left to clean up the record. AI is not a shortcut to safer care. If we allow technology to dictate the speed of nursing care, staffing levels and other clinical decisions, we risk trading real patient care for error and bias prone computer output.

Now in our union survey of more than 2,000 nurses, two thirds said that automated acuity scores from AI Tools these are which attempt to predict how sick a patient is did not match their own professional assessment.

Nurses are not against technology, but CNA nurses refuse to let technology be used to deskill or replace nurses and diminish the quality and safety of patient care without regulatory safeguards.

Healthcare employers, whether driven by profit or under the ever present demand for the increase in efficiency in healthcare, are using algorithmic technologies that interfere with nurses ability to do their jobs and they're gambling with patients lives.

Union nurses urge lawmakers to create a regulatory framework on AI and other algorithmic tech that is based on the precautionary principle that prevents unproven and untested technology from being implemented.

This means putting the burden of proof on AI developers and healthcare employers to demonstrate that their technology is safe, effective and equitable before it's implemented in real world healthcare settings instead of treating patients and nurses as guinea pigs. I want to thank you for your attention to this important subject.

Also say thank you to the people that presented today and we're happy to share more information about CNA nurses experiences with AI with this Committee. Thank you so much.

Good afternoon Mr. Chair. Senator. Nevneet Perrier on behalf of the California School Employees Association just want to reiterate some of the comments already made about technology has its place but should not be used to displace workers especially in the school setting where that human interaction is so important for the development of children. Thank you.

Well, thank you again. I want to thank Senator McNerney for being here. We'll certainly pass on all the information to the other Members of the Committee.

Just a couple quick thoughts since that was since the way you know I think we've we all and I'm going to make sure that this you know we have the start from the premise here that we all see the tremendous potential of new technologies the trend of certainly worked in new technology see the tremendous potential of AI but we also heard a lot of concerning things here.

We don't want to none of us want to have a desk in the workforce or have a job where we're surveilled feeling like we're living in constant surveillance and hearing a lot of concerns about that just now.

But in some of the workplaces I'll just mention quickly I did have a bill that the Governor signed on privacy bill on the extent of some of my work with the delete act that well so data brokers in California will have to disclose if they are collecting information about documentation status.

So I think that's an important part there the piece about hallucinations we hadn't really talked about, but we've seen that certainly in law, and having worked at legal Tech, I follow that pretty closely.

So we've seen the case of hallucinations of cases, cases being made up by AI and all that speaks to, again, why we need humans in the process. I do wonder a little bit.

I'll just wonder aloud in some of the conversations we've had around bias if we can do more testing of some of these systems, testing the output by putting in, say, different resumes of people with different backgrounds, different ethnicities, and see what the input, what the output is, and do that maybe as a simpler way to test some of these systems.

I know some of the approaches that we've had or looked at can be really difficult to implement and can be pretty heavy hand. So that's just something I wonder a lot on as we look forward to legislation that's based on the. I've thought about in terms of testing the output of some of these systems.

Anyway, lots for us to follow up on. And yeah, just with that, I'll leave it at that. Just really appreciate everyone to thank Ian Higgins again from my team. Thank all of our witnesses here today. With that, this meeting of our Select Committee on Economic Development and Technological Innovation is adjourned. Thank you.