We Get AI for Work: Exclusive Interview with Stephanie Yang, Associate General Counsel for ThredUp

December 12, 2024

Some organizations are early adopters of AI, balancing the need to embrace new technology with transparency to stakeholders and adherence to regulatory guidelines.

Transcript

INTRO

Some organizations are early adopters of AI, balancing the need to embrace new technology with transparency to stakeholders and adherence to regulatory guidelines.

On this episode of We get AI for work, we speak with Stephanie Yang, Associate General Counsel at ThredUp and Jackson Lewis alum. Stephanie discusses how AI can help organizations connect with internal and external audiences by being transparent about using the technology. She emphasizes the importance of a cross-functional approach when developing AI usage policies. At ThredUp, Stephanie is part of the organization's internal business team that oversees AI governance.

Today's co-hosts are Eric Felsberg, principal in Jackson Lewis's Long Island office, and Joe Lazzarotti, principal in the firm's Tampa office and co-leaders of the firm's AI Group.

Eric, Joe, and Stephanie, given that companies strive for the best and most efficient experience for the end user, the question on everyone's mind today is: How can organizations adapt AI technology to simultaneously meet their strategic goals and customers' needs, and how does that impact my organization?

CONTENT

Joseph J. Lazzarotti

Principal and Privacy, Data and Cybersecurity Co-Leader

Hello, everyone, and welcome to our latest episode of We get AI at work. My name is Joe Lazzarotti, and I'm here with my colleague and partner Eric Felsberg. This promises to be another interesting episode as we have the pleasure of welcoming and speaking with Stephanie Yang. Stephanie is the associate general counsel at ThredUp. Welcome, Stephanie.

Stephanie Yang
Associate General Counsel, ThredUp

Thanks, Joe. Thanks, Eric.

Eric J. Felsberg
Principal and Artificial Intelligence Co-Leader

It's great to have you here, Stephanie. As Joe mentioned, we're excited for our discussion so let's jump right in. Stephanie, let's start with what is ThredUp and maybe talk a bit about your role with the organization and briefly explain how AI is impacting ThredUp's business.

Yang

Sure. ThredUp is one of the largest online secondhand platforms, particularly for women and children. The organization's mission is to inspire the world to always think secondhand first before they make a purchase decision. If you go onto our website at any point in time, you can see several million listings of accessories, clothing, things of that nature. So, that's the business model.

To jump into today's topic, AI, we are probably one of the earlier adopters of AI. We began looking at how to present a better product experience for the customers when it comes to shopping - back in 2023, shortly after ChatGPT became publicly available.

I started off at ThredUp as the employment law litigator because that was my role. I was at Jackson Lewis - proud Jackson Lewis alum - and ThredUp was a client back when I was in the San Francisco office. And so, I kind of went in-house with that role in mind. The role kind of expanded over time, initially into privacy and product, and then eventually into today, which is entirely cross-functional. Anything business operation related, ranging from product marketing, engineering to now AI, I kind of partner across the organization with our internal business partners.

Felsberg

You mentioned it, but what prompted ThredUp, if you can speak to it, to jump into this AI world? Was it a customer-experience issue? Was it something different? We'd love to hear a little bit more about that, if you could.

Yang

Our CEO and, in general, our executive team are very technology oriented. So anytime there is a new technology that promises to help customers have a better shopping experience, they are always all ears. And so, when ChatGPT first came out, there was, of course, a lot of thinking on how we can use a tool like this or maybe similar to open AI to help with some of the customer pain points.

So, for example, if you went on ThredUp.com pre-middle of 2023, you would notice that the search functionality is there, but it's not always working. So as an example, maybe you type in "Christmas sweaters." In the old version of the search functionality, it's probably going to pull up sweaters that potentially have the word Christmas on them, but not all the sorts of red or green sweaters that you typically associate with Christmas.

Right now, as you go into the search functionality, it's an improved functionality. So, if you search "Christmas sweaters" now, you will get much more relevant results. And that's just one example of what sort of prompted ThredUp as an organization to want to adopt AI to help with that experience.

Lazzarotti

Does that include even ugly Christmas sweaters?

Yang

Yes, actually, I'm glad you asked. If you go on there and search, it will turn up relevant results. Whereas in the past, you could get something kind of strange or occasionally you'll get "Sorry, we don't have anything," when in fact, it's there. It's just that the old search engine didn't know what you were looking for.

Lazzarotti

That's pretty wild. So, another topic I think a lot of people are interested about is transparency and thinking about regulations developing - whether it's for customers or on the employment side - is how companies are using it. And so, how do you balance that? How do you communicate about what you're doing and think about how much or how little and how? Because a lot of this is complex. People don't really know about what really is happening with AI. So maybe talk to us a little bit about that concept.

Yang

Yes. That's a great question. The favorite lawyer answer is "it depends on the circumstance." On the one hand, we as an organization always pride ourselves on transparency. We want our customers to know what we are doing. On the other hand, of course, there's the concern that with new technology, depending on how the disclosure is phrased, customers may be scared away from using it thinking that it's something that it's not.

For example, anecdotally, we've heard from customers initially when AI was adopted - they were like "What is this? I don't understand the disclosure. Maybe I just won't use it." And once it was explained to them - "This is utilizing OpenAI's technology to help you search for things that you may be looking for instead of poking around on the website among several million items." - then the reaction is more "I use ChatGPT all the time. I know OpenAI, this is totally fine." Even though it's really the same tool explained differently and we've anecdotally gotten different types of reactions.

So, trying to balance that, the balancing of transparency, and making sure that from a regulatory perspective we are disclosing because, of course, we are aware that the FTC wants AI use to be disclosed. The SEC is starting to chime in as well for publicly traded companies, and ThredUp is publicly traded.

To balance all that, we try to take a risk-based approach, anyway. So, in the sense that we try to see what is the tool and what's the best way of making sure that customers are aware that there's some sort of generative AI power behind it but without getting so much into the weeds that customers get scared away or maybe think the tool is doing more than what it's doing. As an example, if you go onto our site and as a first-time user, you'll see certain links in different areas and the link takes you to the privacy policy where it talks about the use of AI.

Lazzarotti

Interesting. You mentioned ThredUp as an early adopter of AI. Is that because customers more and more are being exposed to it, they tend to be more willing to understand it and are able to appreciate its value and are okay with using it. Is that right?

Yang

Yes, absolutely right. As the rest of the market starts to adopt AI, it's almost becoming now a customer expectation. So, for example, if you go on ThredUp, there's also the outfit suggestion. It's an open AI tool that's embedded in there where you can chat and you can tell the tool "I'm looking to attend a winter wedding. What should I wear?" It'll come up with some suggestions and then you can tell it "I want more red or less color or longer, shorter." Whatever it is, it'll make suggestions. And it's interesting because customers love it and they also at the same time kind of expect it because there are also other e-commerce companies and retailers out there that are adopting similar tools. And so, to remain competitive in the marketplace, we're probably going to see more and more of it going into 2025.

Felsberg

That is fascinating, Stephanie. One of the things that Joe and I spend a great deal of time discussing is the issue of governance as it relates to AI usage. We'd love to get your thoughts around the governance of AI. Should we have a policy? Should we not have a policy? If we have a policy, what are the contents of that policy? Should we have a core group within the organization that spends its time monitoring not only the AI usage, but maybe developing technologies, developments in the law or any of that? We would really love to hear your thoughts around this issue of AI governance as part of your business.

Yang

I've heard from peers and it sounds like different companies take different approaches. What we do is the following: We try to have a cross-functional approach because at the end of the day to adopt AI, it's not just a legal issue or even an engineering issue. It impacts product, marketing, everybody across the organization. Accordingly, we want to take a cross-functional approach when it comes to AI governance. Internally, we have a somewhat informal AI task force, for lack of a better term, where we have leaders from each organization. So legal, engineering, product, marketing, IT, privacy and security, and operations - they're all in it. We meet async or on as-needed basis, but we are in constant communication. And this way we know which team is thinking of adopting what AI tool and we have an informal repository of what we are doing across the organization.

Felsberg

So, if I were an employee and I hear about this new AI technology that I'm excited about and think the organization would be excited about it, is there an opportunity that I could bring that forward and say "Hey, I found this thing. I think the business could benefit from it. Let's use it."? If you get one of those requests, how do you deal with that?

Yang

Absolutely. We kind of have a ticketing system for this type of inquiry. Or if people don't know about it, they could just approach me or the People Team or whomever or their own leader. Once we get that information, we have the person submit a ticket. And then once I see the ticket or the IT team sees the ticket, they will, depending on what the tool is, cross assign it to the functional leader who's in charge of AI on that team.

So as a random example, if this is like a People Team tool - somebody's like "Hey, we should really look into such and such tool to help with the interview process" - and that's a tool that we're looking at, then the People Team leader and People Operations, as well as myself, will be looped in. We would kind of go look at the tool, understand what it's doing, figure out if there's sort of an enterprise subscription level that is more private than the consumer-facing version. And then, of course, we also vet any sort of proposed contract and terms and things of that nature. So that's kind of how we approach this type of process.

Lazzarotti

Along those lines, Stephanie, you mentioned you have a cross-functional team, which to me sounds like it makes a lot of sense. In terms of the folks on the team and their different areas of focus, how are you managing each different discipline, keeping up with the regulatory developments that we're seeing that may affect maybe a different group differently?

For example, on the HR side, we're seeing stuff from the EEOC and some of the other federal agencies, obviously also at the state level with the New York City law and now we have Colorado, that may have different impacts from an employer perspective than they do perhaps from a consumer perspective. So how do you navigate? What do you do in that group to navigate those issues and stay up to date with a lot of the fast-paced developments that have been going on?

Yang

That's a great question. I try to subscribe to three or four, Jackson Lewis included, law firm publications on this so that on a weekly basis I'm helping the team cross check if there's anything proposed that if passed would impact the business. And then of course, as you know, certain states, California included, tend to be in the forefront. So, I watch those states more closely than others. And then, if we see something that's potentially going to affect, let's say, for example, any sort of consumer-facing websites, which is the core of our business, then I try to have a discussion with either the product team or the marketing team or the engineer or sometimes all three and try to see what solutions we should start to think about in case those changes come into effect. If it's something that's impacting employment, so impacting employment hiring practices, et cetera, then of course, in that situation, I loop in our People Team and we talk through it and we figure out what's the best solution.

It's kind of tough because as you mentioned, these regulations have different angles and different impacts depending on the business function. Personally, I always try to take a risk-based approach of "Okay, how high of a risk do I think this situation is?" Then we try to prioritize accordingly because, to your point, it's possible in any given week or month, there are multiple regulations from multiple agencies coming out. And sometimes it can feel impossible to try to comply with all of them all at once, given all the competing business priorities. And so, we try to slot it in based on the relative risk and impact on the business or sometimes how long do we think it's going to take to build up to that level?

Lazzarotti

Certainly a lot to track.

Felsberg

Stephanie, thanks again for the discussion. Before we let you go, Joe and I have been asking all of our guests to leave us with their top three takeaways as they apply to the use of AI in your business. What are your top three?

Yang

Great question.

1. Number one would be, for me, at least, regardless of what we do, I always want to know what exactly our data is being used for and in what format. And so, when it comes to vetting vendors, especially, because nowadays, different teams discover new and potentially valuable AI tools all the time. And so how do you figure out what's too risky and what's an acceptable level of risk. So that tends to be how I think about it. So that's the number one takeaway: If all else fails, try to look at how and for what they're using the data.

1. Number two, if it's a potentially higher risk situation or maybe the types of data that are being transmitted are a little more sensitive, then I would want to look at what are our indemnification options in the worst-case scenario. As lawyers, we always try to think of worst-case scenarios. And so that's one way that I tried to think when I looked at these AI vendor contracts.

1. Number three is I always want to still take a practical approach - of course, always with compliance in mind. And the reason I say that is because while I want us to be compliant and profitable, I also don't want the legal function within the organization to be seen as the one that's always saying no, right? And so, I want to balance that need to stay compliant and not take on excessive levels of risk with sort of internal trust building to ensure that the team continues to feel comfortable coming to me or anyone else on the legal team with these sorts of inquiries. And because you don't know what you don't know. So. by continuing to foster that, I think that's how I and others in the legal function can continue to stay in the loop on topics such as AI.

Those are the top three that come to mind immediately.

Felsberg

That's good. If you have more, we'll take four and five, but three is good. Three is good. We appreciate it. Thank you for that.

Lazzarotti

That's awesome, Stephanie. Thank you so much again. This has been a really great discussion. I think it's going to be really helpful for our listeners. Certainly, we always hope that they find it helpful, and I can't see how they wouldn't in this case.

If any of you have any questions or would like us to cover a certain AI-related topic, please do so by reaching out to us at [email protected].

Again, Stephanie, thank you so much. Eric, always a pleasure presenting. Thanks to the listeners for listening.

Yang

Thank you both.

OUTRO

Thank you for joining us on We get work™. Please tune into our next program where we will continue to tell you not only what's legal, but what is effective. We get work™ is available to stream and subscribe to on Apple Podcasts, Libsyn, SoundCloud, Spotify and YouTube. For more information on today's topic, our presenters and other Jackson Lewis resources, visit jacksonlewis.com.

As a reminder, this material is provided for informational purposes only. It is not intended to constitute legal advice, nor does it create a client-lawyer relationship between Jackson Lewis and any recipient.