.jpg?cq=50&mw=1180&cpy=Center)
Navigate the intricate nexus of artificial intelligence and the legal domain with Professor James Grimmelmann of Cornell Tech and Cornell Law School and host Carmem Silva. An esteemed authority on the interplay between technology and law, Professor Grimmelmann’s journey from a programmer at Microsoft to clerking in federal courts provides a unique perspective on the legal challenges and opportunities presented by AI. Join us for a thought-provoking journey into the legal dimensions of the AI revolution.
This transcript has been edited for clarity and length.
Zoë Badger: Hello, and welcome to the McKinsey Legal Podcast. I’m your host, Zoë Badger, Associate General Counsel at McKinsey. This podcast is for legal and business professionals, where we’ll explore legal issues on innovation and creativity that matter to you and the world. We will feature interviews with global legal and business leaders to learn more about the ways to unlock their innovative and creative capabilities. We’ll highlight stories on innovators in the legal and business world through a lawyer’s perspective, learn from them, and explore how to amplify our own professional trajectory with insights from our guests.
In this episode, we delve into the fascinating world of artificial intelligence and its intersection with the legal realm. We’re joined by Professor James Grimmelman of Cornell Tech and Cornell Law School, a distinguished expert at the crossroads of technology and law. With a journey that spans from programming at Microsoft to clerking in federal courts in the United States, Professor Grimmelman provides an insightful exploration into the legal complexities and implications of AI. From discussing privacy concerns and the AI supply chain, to the transformative impact of AI on the legal practice, this episode explores how generative AI and law intertwine, shaping our digital future and legal landscape. Join us as we navigate these uncharted waters and uncover the legal intricacies of the AI era.
Carmem Silva: Hello. I’m Carmem Silva, your Brazilian-born and New York-based Contracts Counsel. Today, we have the privilege of speaking with Professor James Grimmelman of Cornell Tech and Cornell Law School. He’s not only a leading voice in digital copyright and internet law, authoring many scholarly articles, but he also translates the complex dance between code and law into insights that shape our digital future. And you should check his thought provoking paper “Talking About AI Generation”. James, great to have you.
Professor James Grimmelmann: It’s great to be here. Thanks for having me.
Exploring GenAI’s Privacy and Copyright Legal Frontier
Carmem Silva: So, James, in the context where AI chatbots are being integrated into widely used digital platforms and services, how do we address the potential legal shifts? Could you particularly address privacy concerns that these integrations might bring?
Professor James Grimmelmann: I think that one of the amazing and terrifying things about generative AI is that it touches on absolutely everything. People are using it for illustrations, to write business pitches, to prepare elementary school lesson plans. It really can do anything. And because of that, the legal issues could really touch on anything that humans do and use computers for. So, you could really throw a dart at any legal issue, and there are going to be generative AIs in there very soon.
When you think about something like privacy, we have at least the advantage of having had some time to think about what these issues are like. I find it helpful to think about generative AI for privacy concerns as being very much like an incredibly good search engine. That search engines confined by exact keyword matching can turn up personal information on people, but because generative AIs, especially things like ChatGPT, can be conversational, they can really help somebody refine their search. And so, since they’re trained on pretty much the entire web, it’s basically the same kinds of privacy dangers that we’ve seen from Google search. I would not be surprised to see the EU start to try to apply the right to be forgotten and GDPR regulations to generative AI. And part of the amazing challenge here for AI developers is that we don’t technically know how to do that yet, that on a technical level, these models work so differently from existing kinds of databases, that that’s a big unsolved technical problem.
Carmem Silva: Can we talk a little bit about copyrights? I am really interested in your paper on gen AI supply chain, the one I mentioned earlier in this episode, and would be great if you could share with us some of the complex aspects of the supply chain. I’m pretty sure our listeners would love to know what are the copyright issues that we, as legal professionals, should be extra aware of.
Professor James Grimmelmann: By the generative AI supply chain, we mean the whole sequence of products and companies that starts with creating training data, then assembling it into big data sets, training models on it, fine-tuning those models for specific applications or aligning them with specific goals, and then putting those into actual deployed products that users have. Every stage of this feeds back into earlier ones, since you might need new training data or have to retrain your models based on the feedback you get further down. This makes these supply chains as complicated as those for traditional products, meaning copyright issues aren’t localized to any one point.
It’s not just when OpenAI releases ChatGPT to the public that copyright issues could arise; lawsuits are now targeting actors at all different levels, from those supplying data to those training base models, to those providing applications to users. What’s unsettling from a copyright perspective is that there are copies at every stage of the supply chain. Starting from copyrighted works like books and web pages and ending with creative work suggests that every stage touches copyright.
Right now, there are many lawsuits against various companies, and the initial rulings have been more favorable for the AI companies than for the plaintiffs. The courts seem inclined to focus on specific infringing outputs rather than treating models or AI systems as infringing. We’re at preliminary decisions from district courts, but that could change.
For now, people working seriously with AI need to be very confident they have a strong transformative fair use case for their application, make sure they’re not showing users copyrighted output, or ensure they’re properly protected by some other actor in the supply chain. The indemnification that Microsoft and others offer to their customers—if you use our products and follow the guardrails, we’ll defend you against copyright lawsuits—is the kind of assurance you might want for your clients.
AI’s Ethical Quagmire
Carmem Silva: Thank you for your great tips. I’ll remember them for my next agreement review for sure. I’m also curious about your thoughts on the recent suggestion to pause AI development. Especially with this movement of influential people and a couple of CEOs endorsing the idea that we are not ready for gen AI. How do you view this? What are your predictions for the future of this technology?
Professor James Grimmelmann: I think from a 10,000-foot view, a pause would be beneficial. We’ve navigated through thousands of years of history without general-purpose generative AIs. If waiting a few more years for the development of really powerful ones means we can approach them more slowly and with better understanding, that’d be ideal. It’d give us much-needed breathing room. However, I don’t see it happening. The business drives of the AI industry, highlighted by the OpenAI power dynamics, are pushing hard in the opposite direction. There are too many stakeholders with significant resources aiming to accelerate progress as quickly as possible.
AI as Creator or Technology?
Carmem Silva: James, we have seen so many headlines about how AI scripted outputs have been shaking Hollywood. So, how do you see the law adapting to this type of AI that blurs the line between the creator and technology? And what do lawyers who represent creators or tech companies need to get ready for in this kind of situations.
Professor James Grimmelmann: I think we have to fundamentally deal with the agency question. Who is responsible for using the outputs of a generative AI? Or for causing it to make a particular output. And sometimes that’s the user who is really intent on a particular task and is trying to push this tool in every direction they can to accomplish that task.
In other cases, it’s more clearly the AI company that provided the system to the user. The user just wants some art. And if the AI then produces something that looks very much like a scene from a Disney movie. It’s really maybe more on the company than the user. In other cases, these are almost inadvertent properties that come out of training generative AIs on a lot of data and unexpected interactions, and there it’s harder to pin down responsibility.
But at the end of the day, what we’re really arguing about is when things go wrong, whose responsibility is it? The kinds of things that can go wrong, at least so far, are very similar to the kinds of things that have been going wrong in law for a long, long time.
Can Generative AI Transform the Legal Profession?
Carmem Silva: I’m not sure if you see it the same way, but from what I’ve been noticing, it’s not just the fear of the new gen AI regulations or the lack of them that is bothering most lawyers. But many of them are actually more worried about losing their jobs to this new technology. And even if they’re not fearing for their jobs to that extreme, they’re definitely concerned about how gen AI is going to change their daily work. So, what’s your take? Do you think that gen AI is going to shake up how lawyers and law firms operate?
Professor James Grimmelmann: I think it will. Generative AI is a classic disruptive technology because it started off being clearly worse than human creativity for almost everything. But there are a few kinds of repetitive text production tasks where it was better, and it got adopted there, and people are training it, and it’s improving. And as it improves, it becomes a serious competitor to all kinds of other tools and human effort. for different kinds of work. It’s not at the level yet where it seriously is going to shape legal practice, but it’s going to be there possibly soon. Right now, the problems of hallucination are too strong to really rely on it for doing legal research or writing. You can see that generative AI tools will be good enough if you can get some of the quality controls under issue.
I think lawyers right now need to be paying attention to these tools. Playing around with them even if just for entertainment, to understand what they’re good at and what they’re not. So that if something does change and it improves dramatically, they’ll be prepared for thinking about how to integrate it into their practice.
Carmem Silva: I think many lawyers are already exploring AI tools and starting to recognize some of the big names in the gen AI legal area. But we know there are still plenty of lawyers out there who are cautious when you talk about gen AI and they seem to prefer getting help from another person instead of relying on technology. So, it’s clear this shift is going to change how we do things in legal services, just like you said.
We’ve covered some really interesting topics today. So a big thank you for that, James. Your insights were key to our conversation. But before we finish, I want to lighten things up a bit while we’re still keeping it insightful.
Lightning round
Carmem Silva: Let’s dive into our lightning round, a quick session where I’ll ask you three fast questions about innovation and creativity. Just go with the first thing that comes to your mind. Okay?
The first one is: could you share something you’ve recently read, watched, or listened to about innovation or creativity that resonated with you? And what about it inspired you?
Professor James Grimmelmann: I thought that David Brassoud’s math book, A Radical Approach to Real Analysis, was absolutely eye opening. It’s a history of calculus in the 19th century that’s kind of a calculus textbook but it’s told historically, and quite honest, everything in it makes so much more sense. When you see the problems with the people at the time we’re trying to grapple with, you see their attempts to solve real physical problems in the world for engineering, and you also see the blind alleys they went down that were not quite right. And after that, all of the rigorous, boring high school calculus made a lot more sense. Like, “oh yeah, this is how we got to where we are today.”
Carmem Silva: I’ll have to check that book then because math was definitely not my favorite subject in high school. And the second question is: we all hit those creative blocks, and do you have a go to, for example, some quick fix strategy or habit that helps you overcome a creative slump?
Professor James Grimmelmann: I go and I play a video game. It’s usually nothing super deep. I’m playing a mountain climbing game right now called Jusant. It’s very enjoyable. It’s just something very different and it really does help me clear my head.
Carmem Silva: I totally get it. I’m also a big fan of video games in general. And finally, when I say the word innovation, what’s the very first word or idea that pops into your mind?
Professor James Grimmelmann: Innovation. Creativity. I know that’s really boring, but that’s like the first word that came to mind.
Carmem Silva: Creativity makes sense. I cannot think about innovation without creativity.
And that is all for now. James, thank you for sharing your knowledge with us. Now I have a lot to think about. To our listeners, thank you for joining us on the McKinsey Legal Podcast. If you enjoyed our conversation, please subscribe. Stay curious. Stay informed. and continue to bridge the gap between innovation and law. Until next time.
Zoë Badger: Thank you for listening to this episode of the McKinsey Legal Podcast on mastering legal challenges in the AI era. Join us for the next episode where we explore legal issues on innovation and creativity that matter to you in the world.
This episode is a production of the McKinsey Legal Department, was produced by Stephanie Spangler. Co-hosted, written, and edited by Carmem Silva. Original music is by David Shaporov.