The following podcast contains advertising.

To access an ad-free version of the LawFair podcast,

become a material supporter of LawFair at patreon.com slash law fair.

That's patreon.com slash law fair.

Also, check out LawFair's other podcast offerings,

rational security, chatter, law fair no bull, and the aftermath.

What it could do is it could actually stitch it together in such a way that it produces something that is word for word exactly what's in its training corpus.

So in other words, what chat GBT is at some level, if we're being very formalistic about it, chat GBT is the author, because it's constructed this, but the exact same information

word for word lives in its training data.

And I think that courts would be reluctant for both kind of substantive and formalistic reasons to impose liability there.

The substantive one is that actually it's not really engaging in any greater harm than already exists in the training data, right?

That sort of bad information is in the training data.

And the formalistic one is that we actually are going to have a good deal of difficulty proving that chat GBT, at least its current versions,

created that piece by piece rather than just pulling a wholesale from its sort of training data.

I'm Alan Rosenstein, associate professor of law at the University of Minnesota, and senior editor at LawFair.

And this is the LawFair podcast for June 23rd, 2023.

There's large language models like chat GBT play an increasingly important role in our society. There will no doubt be examples of them causing harm.

Lawsuits have already been filed in cases where LLMs have made false statements about people, but what about run of the mill negligence cases?

What happens when an LLM provides faulty medical advice or causes extreme emotional distress?

A forthcoming symposium in the Journal of Free Speech Law tackles these questions, and I spoke with three of the symposium's contributors at the University of Arizona and the University of Florida.

Law professors Jane BamBauer and Derek BamBauer and computer scientist Meehai Sardinou.

Jane's paper focuses on what it means for an LLM to breach its studio of care, and Derek and Meehai explore under what conditions the output of LLMs may be shielded from liability by that all-important internet statute, Section 230.

...