Lately, a surprising number of people have asked my thoughts on the intersection of law and generative AI (GenAI). Unfortunately, I never really have a great response. I have lots of thoughts about AI and the law, but it’s hard to convey them coherently because they intersect in almost every way. But I do think the concept of GenAI Exceptionalism and GenuAIflecting is a great framework. (so sayeth the guy who coined the terms)
Society (especially tech companies and elected officials, but also “AI influencers” and AI fanboys) seem all too happy to grant GenAI special privileges that we either (a) don’t give to humans or (b) only give to humans. When we do that, it’s GenAI Exceptionalism. You’d think this would not be controversial, as most Americans would probably think it’s silly…but if you get enough legal scholars stretching the laws in enough exotic ways, sprinkle in the financial incentives of capitalism, and add a dash of rah rah-ing from the likes of Ethan Mollick and it begins to feel inevitable.
GenuAIflecting, as you probably noticed, is just a play on the word “genuflecting,” meaning to show deference or servility. When Congress, state governments, agencies, and others yield to GenAI companies based on nothing more than speculation (GenAI will invent new types of batteries! GenAI will enable nuclear fission! GenAI will help us live forever!), they are participating in genuAIflecting. I think this is mostly from a near complete lack of AI literacy. As I teach my students, GenAI is far less impressive the better you understand how it works and its limitations. When you see a headline that proclaims “AI will,” “AI might,” or “AI could” you should immediately raise your guard. What those statements really mean is “This is a complete guess,” “It’s just as likely this won’t happen,” and “Just trust us!”
When the president opened federal lands to build massive data centers because, boy, we just gotta do it or else America “loses” the “AI race,” that’s genuAIflecting. When California scrapped an AI safety bill because, boy, it will stifle innovation for reasons GenAI companies can’t quite articulate, that’s genuAIflecting. When governments hear Sam Altman say that OpenAI can’t build a world-saving chatbot without the unauthorized use of copyrighted materials, so the governments make carveouts for GenAI companies, refuse to pass laws protecting copyright holders, or knowingly leave regulatory language ambiguous, that’s genuAIflecting.
And, of course, when we grant GenAI special legal privileges that we either don’t grant for humans or only grant for humans, that is both GenAI Exceptionalism and genuAIflecting.
I’m going to try to walk through all this in a series of posts to hopefully make my arguments clearer and, in a small way (my social media presence is tiny) increase general AI literacy in society. While I don’t promise to always hold myself to this outline, here is my initial thought:
A overview/foundational grounding:
Discuss the state of GenAI litigation (aka lawsuits). Who is involved, what claims are being made, what the stakes are.
Identify the gaps in current litigation. What’s being overlooked, which companies are (unfairly) skating by, how GenAI companies are using he gray areas to their advantages.
The role of AI lawyers. What it means, is it really a thing, the role of ethics in legal practice (haha).
Applied GenAI Exceptionalism
GenAI and the First Amendment
GenAI and contract law
GenAI and copyright law
Where I’m headed next
Two (I think) new legal theories to build on the three GenAI Exceptionalism concepts noted above
Questioning open-source as a near-panacea or even a good idea…or even legal, when it comes to GenAI
Note that this assumes the reader already knows what GenAI is. I don’t really want to have to add a caveat each time, but the general one is this:
When I talk about AI, I mean GenAI. I’ll try to always say GenAI specifically because AI and GenAI are not equivalent (GenAI is a subdivision of AI, just like autonomous vehicles, spellcheck, newsfeeds, image recognition, and so much more).
When I refer to GenAI, I mean the general purpose chatbots made by the likes of OpenAI, Meta, xAI, Anthropic, Microsoft, Google, Ai2, and others. You know, the kind that can interact with users and generate images, which obviously means they are the most consequential technology of all time, justifying all expenses and genuAIflecting. I am not referring to specialized GenAI, like those specifically designed for biology or material sciences.
Throughout this journey I genuinely welcome any feedback or criticisms. If I’m overlooking something important, please correct me. I look forward to diving into all this with you!