Artificial Intelligence

Transcript

“One size does not fit all”-John Pavolotsky

John Pavolotsky
Partner and Co-lead, AI, Privacy & Cybersecurity

If I were to look at my crystal ball on the kind of future of AI compliance, again, I don't recommend kind of burying your head and kind of coming back in six months, but you kind of need to go beyond one, of the headline, and two, kind of understand kind of what space am I in, what space are my customers in, and then what is going to potentially impact my business, and where do I kind of want to spend my time for my compliance effort knowing that, okay, six months from now, things might be vastly different.

“…businesses are vetting this technology more before they actually implement it”-Nathan Morales

Nathan Morales
Partner and Co-lead, AI, Privacy & Cybersecurity

I think one of the biggest trends that we're seeing right now as compared to when sort of the idea of AI first kind of took off a few years ago is, and this is actually very comforting to somebody who's a lawyer in this space, is that there's been a back off of this rapid, rapid adoption of anything that has anything to do with artificial intelligence. I think that the businesses behind artificial intelligence were very, very good at marketing it and I don't even think that they necessarily knew what it was that they were marketing at the time. And so now everybody has had a time to step back, better understand the technology and what's going on, and as a result of that businesses are vetting this technology more before they actually implement it. Which means that just because something says AI on it, that doesn't necessarily mean that the business is going to immediately employ it, that it's going to make things easier. Businesses are now seeing that there's risks, massive risks with this technology. They are adjusting their own behaviors to those risks in a way that I see is actually minimizing it for these businesses. They're being more cautious. They're doing more AI reviews to figure out what exactly the technology is doing before they plug it in.

“Understanding what your business does with data is the first step.”-Nathan Morales

I think that any artificial intelligence data strategy is going to first and foremost have to start with the technology. Because every single one of these technologies, even though they all use the same artificial intelligence label, they actually are doing very, very different things. And so I think any strategy first has to figure out and identify what the technology is and what the technology is doing. Within that, what is the technology doing question, the big next step for any sort of data strategy is to figure out what is this technology doing with data? Do we need to feed this technology with additional information? More often than not, the answer to that is yes, because artificial intelligence ultimately is a vacuum for information and so to make it work properly, you almost certainly have to feed additional information into it. So the second sort of piece of what is this thing doing, is what data are we feeding into this? And then again, what is it going to be doing with all of that data? How is it going to be processing it? Is the information, does the information belong to somebody else? If it belongs to somebody else, how do we get that other person to sign off onto it? So I think initially, the first step for any AI data strategy is going to be what is the technology doing and what is it? And then second, after that, it's what is this technology doing with the information that we're feeding it?

“Disclosure is key in AI ethics.”-John Pavolotsky

John Pavolotsky

In terms of AI ethics, I mean there's several kind of components to it: One, is the disclosure…I call it the disclosure component. If AI is being used, do you need to disclose it? Do you need to disclose that you are using it even if there isn't a law on the books that basically says you have to disclose it kind of under these situations? What I advise clients is “yes.” People need to know that this technology is being used even if there isn't a bill that basically says you have to do it though there are bills out there, there's a new bot disclosure bill in California which kind of builds upon…which basically says that to mean that if you're using a bot, you have to disclose to individuals that the bot is being used. But that said, and this kind of goes to kind of more broadly, there are laws out there right now in terms of, for example, it's where you have to be forthcoming and you have to kind of disclose, you have to be kind of truthful and accurate in terms of kind of whatever technology you're using and then how it's being used. That can be enforced by the Federal Trade Commission, it can be enforced by the state attorney generals under kind of the umbrella of consumer protection. If it is not prohibited here, it might well be prohibited somewhere else and so you look at kind of your different jurisdictions and thinking, okay, there's a decent chance that whatever the law is in a different jurisdiction it is going to make its way over here. A good example would be the EU AI Act. And so yes, if you don't sell in Europe, okay, you could do it here potentially but it probably will make its way over here, one. And two, you have, what I call kind of effectively normative considerations. So even if, and that kind of goes to ethics, so even if there's nothing that says I can't do it, should I? And so that's kind of where the conversations there with clients happens.

“If it’s not prohibited here, it might be prohibited somewhere else”-John Pavolotsky

Nathan Morales

The big question with IP and artificial intelligence is if you are feeding these artificial intelligence models with somebody else's intellectual property, what is it that that artificial intelligence model is doing with that intellectual property again, and is it altering that property in a way that first of all is legal, but second of all, that is turning it into something else that is making a new creation or a new piece of intellectual property, and who does that new intellectual property belong to especially if it's derivative of the works of others? And we don't know the answer to that question. I can tell you that I think that a checklist is extremely important. In fact, when we here at the firm saw how much work was going to start coming in around artificial intelligence, one of the first things that we did was start talking about a checklist for clients. And again, this relates directly to the huge universe of risks that artificial intelligence can get its fingers into. Again, like I said, so we have potential labor and employment risks. We have potential litigation risks. We have potential corporate risks. And one thing that a checklist can do is it can very easily capture that universe for you in a way that is simple and you can just sort of run through it as you are sort of conducting your AI analysis and adoption sort of process, and really just simplify that process. So we use checklists all the time in that world because I think we would just lose out on too much of that stuff if we didn't.