Ok, but Would AI Judges Really Be Any Worse?
Added 2024-12-23 08:59:09 +0000 UTCOA1103 - Is human intelligence necessarily more rational and just than artificial intelligence? How involved should AI be in our law and government? Professor Aziz Huq of the University of Chicago School of Law joins for a fascinating conversation about everything from the “right to a human decision” to the dystopian terrors of Tinder.
“A Right to a Human Decision,” Aziz Huq, Virginia Law Review (2020)
“The Geopolitics of Digital Regulation,” Aziz Huq (2024)
“Chinese scientists develop AI ‘prosecutor' that can press its own charges,” Steven Chen, South China Morning Post (2021)
Check out the OA Linktree for all the places to go and things to do!
Comments
Professor Huq: So, the law is like a really old program made by a bunch of people who aren't around to explain it any more Me, a software developer: Oh. Oh no.
Merk
2025-01-04 05:47:49 +0000 UTCAnd the money is only part of it. You and Professor Huq were right about the difficulties of weeding prejudice out of AI even when you're trying, but if someone like Musk were supplying the code, I'd have no confidence they were even trying not to tip the scales.
Gmork
2024-12-23 22:06:12 +0000 UTCCasey said the same thing when she listened to this, i had no idea how much the state contracts for some of these systems really were!
Matt Cameron
2024-12-23 21:57:16 +0000 UTCGot you there, that is a really good point! Clearly i have a lot more to learn about ai myself, we'll be back on this topic sometime soon i'm sure
Matt Cameron
2024-12-23 21:55:47 +0000 UTCWhat a guest! Please find excuses to have him back on. I'm still mulling over his opening analogy about law and programming. The thought of Judges as these societal debuggers is fascinating.
Jon C
2024-12-23 21:04:22 +0000 UTCGreat episode overall, but the reason I'm being pedantic on this metaphor is that I see the opportunity/danger of AI being that's its great for suggestions/first drafts but disastrous in the context of final decision-making like when lawyers trust it to write their briefs for them. Dating apps just seem particular distinguishable bc nobody ever trusts the apps with final authority of the decision of who matches who.
Gmork
2024-12-23 18:05:25 +0000 UTCas someone who has been using dating sites/apps pretty much continuously on and off throughout their entire history--even well before they were all that useful for non-monogamous purposes--I'm not sure that I entirely agree with either of you. I think the overall point that he was making was something that I have thought about a lot: a part of our lives which has for millenia traditionally been decided for better or worse by community/social ties/geography/class is now being mostly controlled by a couple of tech companies which have no moral or ethical obligations to us whatsoever. AI isn't (yet) really part of the equation (although it does now do things like help you write profiles and even opening messages [ick] and select the most attractive photo) but the model and algorithms behind Tinder/Bumble/Hinge favor, in order: (1) physically attractive people (but when has humanity not done this), (2) people who are willing to pay extra for premium features so that their profiles come up more often, and (3) people who use the apps more frequently so that their profiles come up more often. This makes them by default much more accessible to people with more disposable time and income, and of course being able to search by things like race and class indicators (as OKCupid does) supercharges existing prejudices and denies us the chance to organically meet people who we might not otherwise think we were interested in meeting. I have a *lot* more to say about this if I thought it through a little more but IMO it's a little more complicated than how you have characterized it and probably less complicated than Prof. Huq (who has of course acknowledged here that he has never actually used these apps himself) has characterized it. Really fascinating stuff though, I wish we had an excuse to do a whole episode on it--but maybe when one of these apps is inevitably sued for intentional infliction of emotional distress or whatever?
Matt Cameron
2024-12-23 17:49:04 +0000 UTCI've been listening to the back catalog of Philosophers in Space recently, and coincidentally listened to the Minority Report episode just the other day. Thomas, one of the major themes you and Aaron talked about was how the story (rather than the movie) deals with the question of "even if this tech is perfect, is it ethical/good/etc?" which seems really relevant here. Questions of reliability bother me in the same way "fetal viability" always bothers me in discussions of abortion rights -- it's a standard that, almost by definition, is going to get less useful over time as technology improves, and it lets us ignore the meaningful questions.
Mike Lazzaro
2024-12-23 15:07:06 +0000 UTCTo illustrate why I'm skeptical of the use of A.I. for legal interpretation of language, I wrote a law review article analyzing when claims are "related" for coverage purposes. That has a very nuanced meaning in that context. If you asked ChatGPat what it meant from a review of the internet, literature, or even solely limited to court cases, the result would be useless gibberish.
Gmork
2024-12-23 14:01:36 +0000 UTCI think Professor Huq radically misunderstands what dating apps are or how they operate. Nobody surrenders their judgment to an algorithm as to who to date. They're simply presented with the options of other people using the app and use their own human judgment as to who to reach out to.
Gmork
2024-12-23 13:53:12 +0000 UTCAside from the moral good or efficiency, he's dead wrong that they A.I. will be cheaper. In theory, it *could* be performed more cheaply, but it'll be dependent upon licensing the A.I. from tech bros who designed it for the express purpose of charging as much as they can for its use
Gmork
2024-12-23 13:47:07 +0000 UTCYou guys desperately need to interview Dr. Emily M. Bender. Both of your AI discussions recently came from a faulty understanding of what AI can do today intentionally perpetuated by marketing. (My opinion now) AI is just a new way for corporations and governments to "launder" a decision to pretend they have no responsibility for it. 10 years ago it was "the algorithm", 20 years ago it was "the economy", now it's AI. In reality it's not requiring those who make decisions to do something in the human world to take any responsibility for the effect it has on humans.
Whitty
2024-12-23 11:43:27 +0000 UTCSick, interesting topic, jumping on this one! Also, thank you so much for the episodes around holiday season, I realise how much extra effort they must be to squeeze in xxxxxx
oh good what luck
2024-12-23 09:01:41 +0000 UTC
