David Gerardtoday's (unreleased, likely ~spring 2024) legal opinion article: AI models are likely uncopyrightable (no human hand) and too easy to reverse-engineer for trade secrets to work. So the VC billions are going into stuff they can't own. Which is the funniest possible result.
Bruno Nicoletti@davidgerard They will throw an enormous amount of money at lawyers in an attempt to challenge, along with lobbying to change it. My fear is that those efforts may very well succeed. If you remember Sunak fawning over Musk while being obviously totally out of his depth, I wouldn’t put it past him to bend over backwards for the techbros.
Cykonot
acb@davidgerard They’re unownable and dangerous for anything that’s more than a novelty, but they do an excellent job of flooding the zone with Augean quantities of bullshit, which may be convenient for those who don’t like democratic culture being viable. In other words, maybe it’s not about the blanket but the smallpox?
🐻 John@acb @davidgerard "It's not about the blanket, it's about the smallpox" is a hell of a turn of phrase
Blayne Haggart@davidgerard @davidgerard The thing about IP law and its interpretation is that both change over time to reflect the relative power of the relevant interest groups, favouring either protection/control or dissemination. They don't reflect any consistent objective principles. (Topic of my dissertation and book, Copyfight: The Global Politics of Digital Copyright Reform.)
If it's seen as important enough, either the law or the court's interpretation of the law will change.
SpaceLifeFormSign up with our LLM/AI today!
Soon, we will generate the bestest, coolest, custom NFTs just for you!
Profit!
d@nny "disc@" mcClanahan@davidgerard need this out expeditiously but i'll let the chef cook
Niedliche Nacktschnecke ☑️@davidgerard you own an ai model by owning it, not through copyright?
Or is there an actual business model where you "lend" someone the full model?
Or is there an actual business model where you "lend" someone the full model?
@niedlichenacktschnecke the model is a constructed pile of data created by training. that model is likely not copyrightable.
@davidgerard but arent alk these air applications server based, where the ai model never goes into the customers hands? That's what I'm missing
Jon@davidgerard Sorry, I think you are confusing the available to download models, like on HuggingFace, where your argument has some value with the proprietary and not downloadable models like GPT 4. Good luck trying to reverse engineer GPT 4 from typing in prompts and recording responses. The valuable intellectual property models don’t depend on copyright law to retain their value, they depend on secrecy and security - like the recipe for Coke.
@JonC the dude writing this is pretty sure it's laughably doable, and the closest they have to a moat is access to lots of Azure GPUs
TorontoWill@davidgerard @JonC I was wondering this as well, since one of the characteristics of these bots is that their path from to A to Z is so inscrutable that even the creator usually doesn’t understand how it got there (without painstaking research). But I suppose that’s not the coded part (thus why it’s inscrutable) and the code holding it together is maybe not so complicated.
g-clef@JonC @davidgerard Look for papers on “model leeching”. Copying a model by prompting it millions of times is a thing. It’s a fun question: if two models return the same answer for every prompt (or 90-plus percent of them), are they copies of each other? Even if their internal weights and vocabs are totally different?
@gclef @davidgerard There is a big difference between using a set of prompts and responses (ala https://arxiv.org/pdf/2309.10544.pdf) to train a student model and having a reverse engineered copy of a model like GPT.
Robert Link@davidgerard For sure the legal system that failed to notice shrink-wrap licenses are the very model of "contracts of adhesion" will also come to the aid of the latest round of tech-nonsense.
Anton Leuski@davidgerard what about the models that require human judgments for training? Isn’t that a human hand? Human relevance feedback in OpenAI parlance.
J Wynia@davidgerard which is why Meta, Microsoft and others are giving so many of the models away. They realize the value has to be elsewhere.
Tim Bondy@davidgerard
"This is looking like a whack-a-mole issue. And way more complex than what the average citizens will be able to understand.
"This is looking like a whack-a-mole issue. And way more complex than what the average citizens will be able to understand.
GVF 🇨🇦@davidgerard Oh noze. Dumping a billion into political influence won't help?
Jesse Cohn@davidgerard pardon my legal ignorance if I'm asking a stupid question here, but... wouldn't someone just have to lightly edit an AI creation to add "human hands" for legal purposes?
Eckes 
@davidgerard they can if they only host them or regularly create new ones,, in the end it’s the ecosystem, not the model