Episode 186: AI’s Workplace Opportunities, Challenges

Welcome to the Workplace, a podcast by Cal Chamber. I'm Matthew Roberts, the Labor Law helpline manager and Employment Law council with the California Chamber of Commerce. Hello, listeners. Well, we have a special treat for this episode today. We will be discussing one of 2023's novel and hot topics, artificial intelligence, or what's normally known as AI. Of course, news stories are prevalent this year about things like AI language models and how they might change how humans do things in their personal lives. But there is much more to AI than just the programs that have dominated the news. There are AI applications that are already impacting the workplace and many more that might alter the workplace altogether. To discuss concepts around AI in the workplace, we are very proud to welcome Peter Bittner. Peter is an award winning multimedia journalist, AI consultant and media strategist. Peter's work has been featured in the New York Times, CNN and other major media outlets. A Fulbright scholar and UC Berkeley lecturer in new media, Peter is deeply involved in the intersection of technology and storytelling. Passionate about learning, utilizing and teaching the latest tech tools, Peter is also dedicated to guiding businesses in leveraging artificial intelligence for powerful and impactful narratives. All of that is to say I am very glad to welcome you to the show today. So thank you for joining me, Peter.

Matthew, thanks so much for having me. It's an honor to be here.

Excellent. Well, Peter, we're hearing a lot about AI and its role in the workplace here in California. We're seeing some employees and managers have kind of a built in fear around it. Others are seeing that it can improve productivity and eliminate many repetitive back office type functions. Peter, can you talk a little about how much AI has expanded in workplaces and how companies are currently using AI?

Absolutely. Let me start with a little bit of data about the rapid expansion of AI in the workplace. So Salesforce had a global survey of roughly 11,000 workers that came out earlier this year. And they found that globally, you know, two thirds of managers said that their company is considering or implementing generative AI in one form or another. So this is a massive transformation. This is a disruption within business models for many different industries and it's also a hype cycle. So we need to, you know, be wary of that as well and think critically. I will say that, you know, there is also a huge need that was found in the Salesforce study for training. Nearly 97% of global workers who participated in this research said that they believe businesses should prioritize AI skills in their employee development strategy. That's almost unanimous. So there's a lot of room for employers to get involved in training their workers to use these sophisticated AI tools, which are varied and multifaceted. There's AI for all different types of tasks. Right. That can be kind of a mini superpower for many different types of workers. Right. More workers now in the Salesforce study were more excited about the use of AI in the workplace than worried about it replacing them. Right. We have to acknowledge that there is some uncertainty and some preoccupation by workers of being replaced by some AI. Right. Which is a little dystopian. So we need to acknowledge the elephant in the room, and I encourage employers to do that as well. You know, this is something that is really a next wave of automation, right. It's AI has been implemented and been happening within all sorts of tech companies for years. Right. But it's really this year, 2023, where we've seen kind of a milestone when it comes to generative AI tools. So things like ChatGPT, right, for text generation, things like Mid Journey or Dall E for image generation, and, you know, a huge variety of video generation tools and software out there as well, which are really advancing so rapidly now that they are, you know, leading to some real interest. Right. So ChatGPT gained an incredible following. You know, they reached 100 million users faster than any other app in history. That is until Threads, the Twitter, or should I say X knockoff came out a little bit later in the year, this year. But obviously there's a lot of interest in these tools.

Right, Peter? And, you know, so much of what we've seen or heard around the media, I think, is around AI and the dangers of it. I know some of the program designers around those generative AI programs, as we talked about, have been really careful to be at the forefront of talking about how they will mitigate any kind of perceived dangers around AI. Governments, including ours here in California, has been making resolutions and statements and plans about how they will address AI in the workplace. But honestly, as I'm listening to you talk, I'm a lot more excited about what this can mean for workforces. So I'd rather focus on the positives to start. What do you see as some of the upsides of businesses using AI in general as part of their operations?

Absolutely. Yeah, that's a great question. So there was just another study. Sorry, I'm a university lecturer normally, so I love research. And this study that just came out, again, this one was with Boston Consulting Group in collaboration between Harvard, UPenn and MIT professors that analyzed hundreds of consultants across, you know, 18 different work tasks. They found that those who used AI finished, you know, almost 12% more tasks. Those who use AI also produce 40% higher quality results than those who didn't. And they completed tasks 25% faster. Right. And so these are obviously, you know, early figures here. It's still a little bit rough around the edges. This is a working paper from, from Harvard. Really. And so it has yet to be peer reviewed. But so these figures are really in line with other research that I've seen over the last 12 to 18 months, which show that there's a bump in productivity per worker of roughly 20 to 40%. This is substantial, obviously across an entire organization or company. And businesses should be paying attention to how to best optimize their workflows and operations with these different types of tools.

Yeah, Peter, shifting gears a little bit, you did mention this a little bit about some dystopian fears about AI in general, some more specific fears about AI replacing human workers. I know we've seen labor issues here in California and elsewhere about their work being replaced by AI and some down, you know, some downsides to stuff, copyright issues and things like that. So Peter, really, as far as workflow and operations, what are the major downsides or what I should say the major issues that employers and businesses should be aware of and address as they're looking to implement or expand their use of AI?

Of course. Yeah, that's a great question. So, you know, a lot of employers immediately look at productivity, look at things like cost savings, you know, accuracy and consistency, potentially, you know, ruling out human error. Right. The massive adaptability that these machine learning models, you know, can improve and learn and iterate upon themselves over time. Right. So these are the obvious draws, but the negatives. Right. The risks that you alluded to are very serious as well. So it's certainly a double edged sword. Right. Things like job displacement. Right. For example, really sort of automating routine and repetitive jobs such as admin, call center roles, entry level positions, software engineers. You know, while these things are becoming more possible, it might not be wise to do that for a number of reasons. Right. You still need a human layer for, for almost every single type of AI to be able to apply some actual human judgment and quality control. Right. And it is a little bit alarming, you know, when there are headlines that say, you know, X firm put a hiring freeze on. They cited AI in part due to this. And you know, it's something that I think is going to be difficulty for, for attracting and maintaining pipeline of talent, especially for graduating students. Coming into the job market and retention. So there's also a few other issues here. Right. So obviously, you know, the reliability of these products too. Right. Are famously in question. So there's a lot of, you know, wonkiness and unpredictability in the use of everything from chat GPT to midjourney to advanced language models or, you know, data analysis, analysis ones. Right. And we need to really be sure that there are checks and balances and internal processes for mitigating those. There's also a lot of different risks of, you know, security especially, you know, AI systems are not only in the hands of our businesses, but also adversaries such as cyber, you know, hackers and all sorts of different organized groups. So, you know, it's really important to maintain budget and vigilance when it comes to this. Of course, bias and DEI issues. And a lot of these AI tools are critical, you know, to acknowledge. And one of the biggest risks, you know, for businesses in this environment is making sure that they mitigate for bias and account for DEI issues in rolling out any of these AI systems. Right. If they're not trained properly on representative data sets or implemented correctly, or if there's not the right people in the room to be able to actually, you know, build these products or apply them in your business, they can reinforce societal biases and have tremendously negative impacts. And potentially that could put your business at risk of, you know, legal jeopardy, really. I think we're going to talk more about compliance issues next. I'm excited to dig into that. It's certainly a big area and it's an evolving playing field.

Yeah, Peter, I mean, we absolutely are. And I think the bias reinforcement is really important, and it really hammers home the humans that still need to be around AI, because AI with generative AI, as you've talked about, can create so much stuff, but what it creates just may not be relevant or may, in fact, actually be dangerous for the kind of idea or mission statement or product you're trying to put out. Now, of course, this year we've seen some language from the Federal Equal Employment Opportunity Commission, the EEOC that we've talked about on the show, as well as the California Civil Rights Department about efforts to create rules around AI in the employment context. Now, we did just see the EEOC execute a settlement against an employer for AI practices that ended up unlawfully screening out older workers. Just in general, one of these disparate impact of claims, in addition to those kind of screening and hiring practices, what other legal and compliance issues should employers consider when using AI in the workplace?

That's such a great question. And it's a bit of, you know, a dynamic legal environment here. Obviously, you know, Governor Gavin Newsom in California earlier this month in September issued an executive order that essentially, you know, gave a very serious homework assignment to most major agencies in the state of California. They have 60 days to really do a deep dive into what are the threats, opportunities involved in their specific department. So everyone from CDOT to some of the other agencies are really going to be rolling out some of their own suggested best practices. And there's obviously a lot of legislative talks that are ongoing to this fall. And we're going to have to also keep our eye on some of the national, you know, potential legislation that might be coming with this recent, you know, closed door AI summit, as some are calling it, right. That the Biden administration just hosted earlier this past week. So, you know, it's. It remains to be seen what is coming down the pipeline. And that makes it a bit tricky as a business owner, right, of any size business to determine where to, you know, implement or hold off. But I think, you know, looking at the California Consumer Privacy act is critical. You know, biometric data is something that is really, really important to be in, you know, alignment with and making sure that you're aware of some of the pitfalls of these AI powered customer screening, you know, or potentially hiring tools like HireVue, for example. You know, there's a Illinois lawsuit in the past few months here that has to deal with informed consent during the hiring process. Essentially it's a video interview tool that uses very sophisticated machine learning algorithms to analyze body language. Right. Intonation, all sorts of different, maybe eye movement to assess the confidence and capabilities of potential employees. So, you know, these are things that are now available as products at enterprise levels that obviously have a certain appeal and perhaps a place in the workplace. But it's definitely a very shaky territory right now. Amazon famously scrapped its sort of secret AI recruiting tool that showed a large bias against women in 2018. Right. And this is kind of the backdrop of, you know, some of the current legal issues of recruiting tools that are using AI algorithms. There's a lot of evidence that shows that these types of AI systems reinforce bias and discrimination in screening processes. So, you know, these are some hot legal potatoes to juggle and you really need to be in consultation with your lawyers. I am not a legal expert, I will say and give a disclaimer that this is not legal advice. But I am, yeah, doing my very best as well to stay up to date with the latest in Sacramento.

Yeah, I absolutely piggyback on that too. And what I kind of think about when I hear you talk about the screening tools, the videos that picks up eye movement and body movements and language and how their tone is, I think back to something that gained some popularity not too long ago in hiring practices, which are these, like emotional intelligence tests or these psychological exams, which we didn't have a lot of guidance around originally at the time. But over time, as you've talked about with AI, there's potential for some bias or again, disparate impact on particular protected classes like national origin or race or sex. And you can kind of just use your imagination about what a computer might miss or how a computer might classify a whole group of people on how they look or what their cultural traditions might be and how they address, you know, strangers or formal interviews. So I think it's really important because we've learned those psychological and emotional intelligence tests probably have not enough benefit to outweigh the massive con of this discrimination claim. One thing I wanted to talk about with you that I kind of put a pin in while we've been talking here are these privacy issues, because you didn't mention it. California has a pretty robust Consumer Privacy act at the moment, and we've talked a lot on this show about how that applies to employers. What I'm really interested in from you, Peter, is how are the privacy issues arising from using AI models, meaning sensitive information that we keep track of. Personnel information, of course, is private under the California Constitution. And how can AI kind of open us up to liability for these privacy issues?

Yeah, that's a great question. So AI is really kind of the next level of really using and collecting data at a scale that is previously almost unimaginable. A decade ago. Right. I mean, you could say the 2000 and tens up until, you know, a few years ago was the era of big data when it comes to big business. Right. And now we are seeing with AI the ability to manipulate, analyze, splice and dice, share, integrate, synthesize, you know, these data sets of employees of potential, you know, sales leads and funnels of all types of proprietary data in ways that now creates a ton of possibilities and excitement for increasing productivity and understandings of where the ball is moving and where we should be chasing. But it's also opened up a lot of different questions about the sharing of data between, you know, third party systems, Right. Our vendors, or, you know, enterprise systems that you're using that are AI specific, you know, sharing data with other outside entities. That's a big risk as well, because the scale at which you know and who owns that data? Right? Do you own the data? Whose servers are those on? You know, these are some questions that are more important than ever now in 2023. It pays to read the fine print and ask very critical questions of vendors and to dig into the nitty gritty.

Excellent. Well, Peter, I think we've addressed quite a bit here on the show during the course of this discussion. A lot of the positives, some of the issues, both just practical and compliance, that employers should at least consider and address with counsel and other experts before we just go head first diving into the wild, wild west here as we are. But I just wondered if there's any other issues or perks of AI that we haven't really discussed that business owners, HR professionals or managers should be aware of.

It's a great question. Some of my sort of initial advice is going to be if you're an executive, if you're a business owner, you must learn the tools that are related to your industry and your operations. It is critical, do not get left behind. Yes, there is hype cycle here when it comes to Silicon Valley and, you know, this recent sort of venture capital environment in which we find ourselves, right. There are reasons to be, you know, critical, to be wary. Don't rush into any decisions, but you need to do your homework and explore these tools now before your competition does. Seriously. It is a big deal. These are critical issues that you need to understand and you need to bring in, you know, your board, you need to bring in your fiduciaries, you need to bring in your key stakeholders, your business partners, potentially, you know, some of your other key parties here to strategize and figure out where the opportunities and threats are. Do a SWOT analysis for your business, right. And bring your employees along the journey as much as you can because they may be learning, you know, faster than you and be able to step up and become a leader in this space within your organization. This is an evolving, you know, enterprise area here. So you need to be constantly educating yourself and dedicating, you know, HR hours to trainings once you have established best practices and, you know, found some reliable tools and ways to increase operational efficiencies. It's really, really important, you know, to have ongoing trainings because the productivity gains, right. Are really, really important to not let slip away and, you know, get behind competitors in this area. That said, don't let the fear and hype lead you into jumping into decisions that are impulsive. Right. Make sure to have ethical considerations, responsible use at the forefront. Think about your core values, your mission as an organization. Make sure that you're using and envisioning AI in a way that aligns with our, you know, societal values and norms as well as your specific mission and values as a company. And don't forget about your customer. Right. You need to be transparent with them as well. And you know, hire a lawyer. There's not a lot that, you know, are going to maybe want to be on the record as this, you know, legal environment is constantly changing. But as soon as those laws come out, you know, in the state of California, nationally, it's going to be critical to get legal review as well as soon as possible in this process.

Yeah. And I love that, Peter, because, you know, anyone who's listened to me talks know if you're dealing with third parties with the organization, especially in the employment context, but really in any context that doing your homework on those third parties that are going to do some important employment related function. For example, if you're going to hire a third party to do your criminal background checks, if you're going to hire a third party to administer your legally required leaves of absence, if they mess it up, you as the employer are still the one who's on the hook, right? No one is suing your third party. They're suing you. So I think that lesson is really well taken and extending here to the AI because this is a big, broad, brave new world here. Just want to leave listeners with that again, as you do your homework. Just remember, ultimately, whoever you're comfortable with, be really comfortable with them in your roles and what they're going to do for you because it's always going to come back to us as the employer. Peter, this has been a really wonderful discussion on AI in the workplace. I really appreciate you taking the time out to talk with me about these issues. Thank you so much for joining me today.

My pleasure, Matthew. Yeah. And for those interested in, you know, getting updates and following along with some of the emerging, you know, tools as well as the evolving landscape in the workplace relating to AI, they can check out my regular newsletter. It comes out weekly. It is called theupgrade AI where they can learn more. I'm constantly educating myself and putting together, you know, a list of top stories related to the workplace and hopefully it'll be a good resource for some of your listeners.

It sounds fantastic. Listeners, grab ahold of your experts wherever you need them. So thank you again, Peter. And thank you listeners for joining this discussion on the workplace. Please comment, share and subscribe to Cal Chambers podcast by visiting calchamber.com.