Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

OpenAI departures: Why can’t former employees talk?

vox.com

Post reply on HN
Print view View original

421–430 of 1001 posts

Re: OpenAI departures: Why can’t former employees talk?

#421

Post by KoolKat23 »

kmeisthax wrote: post #405
bluefirebrand wrote: post #339

Earlier quoted context omitted.

Seems absurd that somehow the scale being massive makes it better somehow You would think having a massive scale just means it has infringed even more copyrights, and therefore should be in even more hot water

So, the law has this concept of 'de minimus' infringement, where if you take a very small amount - like, way smaller than even a fair use - the courts don't care. If you're taking a handful of word probabilities from every book ever written, then the portion taken from each work is very, very low, so courts aren't likely to care. If you're only training on a handful of works then you're taking more from them, meaning…

You don't even need to go this far.

The word-probabilities are transformative use, a form of fair use and aren't an issue.

The specific output at each point in time is what would be judged to be fair use or copyright infringing.

I'd argue the user would be responsible for ensuring they're not infringing by using the output in a copyright infringing manner i.e. for profit, as they've fed certain inputs into the model which led to the output. In the same way you can't sue Microsoft for someone typing up copyrighted works into Microsoft Word and then distributing for profit.

De minimus is still helpful here, not all infringments are noteworthy.

Re: OpenAI departures: Why can’t former employees talk?

#422

Post by sratner »

mvdtnz wrote: post #367
romwell wrote: post #345

Earlier quoted context omitted.

Unless the terms of the NDA are provided upfront, that sounds sketch AF. "I agree to follow unspecified terms in perpetuity, or return the pay I already earned" doesn't vibe with labor laws. And if those NDA terms were already in the contract, there would be no need to sign them upon exit.

> And if those NDA terms were already in the contract, there would be no need to sign them upon exit. If the NDA terms were agreed in an employment contract they would no longer be valid upon termination of that contract.

Plenty of contracts have survivorship clauses. In particular, non-disclosure clauses and IP rights are the ones to most commonly survive termination.

Re: OpenAI departures: Why can’t former employees talk?

#423

Post by PontifexMinimus »

NewJazz wrote: post #344
bluefirebrand wrote: post #339

Earlier quoted context omitted.

Seems absurd that somehow the scale being massive makes it better somehow You would think having a massive scale just means it has infringed even more copyrights, and therefore should be in even more hot water

My US history teacher taught me something important. He said that if you are going to steal and don't want to get in trouble, steal a whole lot.

Copying one person is plagarism. Copying lots of people is research.

Re: OpenAI departures: Why can’t former employees talk?

#424

Post by kmeisthax »

staticautomatic wrote: post #368
eru wrote: post #233

Earlier quoted context omitted.

> What's the consideration for this contract? Consideration is almost meaningless as an obstacle here. They can give the other party a peppercorn, and that would be enough to count as consideration. https://en.wikipedia.org/wiki/Peppercorn_(law) There might be other legal challenges here, but 'consideration' is unlikely to be one of them. Unless OpenAI has idiots for lawyers.

Ok but peppercorn or not, what’s the consideration?

"I'll pay you a dollar to shut up"

"Deal"

Re: OpenAI departures: Why can’t former employees talk?

#425

Post by KoolKat23 »

mwigdahl wrote: post #290

The best approach to circumventing the nondisclosure agreement is for the affected employees to get together, write out everything they want to say about OpenAI, train an LLM on that text, and then release it. Based on these companies' arguments that copyrighted material is not actually reproduced by these models, and that any seemingly-infringing use is the responsibility of the user of the model rather than those w…

Lol this would be a great performative piece. Although not so sure it'd stand up to scrutiny. Openai could probably take them to court on the grounds of disclosure of trade secrets or something like that and force them to reveal its training data and thus potentially revealing its source.

Re: OpenAI departures: Why can’t former employees talk?

#426

Post by ipaddr »

riehwvfbk wrote: post #397
alexpetralia wrote: post #282

If the original agreement offered equity that vests, then suddenly another future agreement can potentially revoke that vested equity? It makes no sense unless somehow additional conditions were attached to the vested equity in the original agreement.

And almost all equity agreements do exactly that - give the company right of repurchase. If you've ever signed one, go re-read it. You'll likely see that clause right there in black and white.

For companies unlisted on stock exchanges the options are then worthless.

These were profit sharing units vs options.

Re: OpenAI departures: Why can’t former employees talk?

#427

Post by eternauta3k »

yumraj wrote: post #225

Compared to what seemed like their original charter, with non-profit structure and all, now it seems like a rather poisonous place. They will have many successes in the short run, but, their long run future suddenly looks a little murky.

It could work like academia or finance: poisonous environment (it is said), but ambitious enough people still go in to try their luck.

Re: OpenAI departures: Why can’t former employees talk?

#428

Post by windowsrookie »

jimnotgym wrote: post #398

>the company will succeed at developing AI systems that make most human labor obsolete. Hmmmn. Most of the humans where I work do things physically with their hands. I don't see what AI will achieve in their area. Can AI paint the walls in my house, fix the boiler and swap out the rotten windows? If so I think a subscription to chat GPT is very reasonably priced!

Obviously if your job requires blue-collar style manual labor, no it's likely not going to be replaced anytime soon.

But if your job is mostly sitting at a computer, I would be a bit worried.

Re: OpenAI departures: Why can’t former employees talk?

#429

Post by solidasparagus »

jay-barronville wrote: post #197
adamtaylor_13 wrote: post #191

Earlier quoted context omitted.

Reading that thread it’s really interesting to me. I see how far we’ve come in a short couple of years. But I still can’t grasp how we’ll achieve AGI within any reasonable amount of time. It just seems like we’re missing some really critical… something… Idk. Folks much smarter than I seem worried so maybe I should be too but it just seems like such a long shot.

When it comes to AI, as a rule, you should assume that whatever has been made public by a company like OpenAI is AT LEAST 6 months behind what they’ve accomplished internally. At least. So yes, the insiders very likely know a thing or two that the rest of us don’t.

But you also have to remember that the pursuit of AGI is a vital story behind things like fundraising, hiring, influencing politicians, being able to leave and raise large amounts of money for your next endeavor, etc.

If you've been working on AI, you've seen everything go up and to the right for a while - who really benefits from pointing out that a slowdown is occurring? Who is incentivized to talk about how the benefits from scaling are slowing down or the publicly available internet-scale corpuses are running out? Not anyone who trains models and needs compute, I can tell you that much. And not anyone who has a financial interest in these companies either.

Re: OpenAI departures: Why can’t former employees talk?

#430

Post by ben_w »

wtallis wrote: post #418
kmeisthax wrote: post #405

Earlier quoted context omitted.

So, the law has this concept of 'de minimus' infringement, where if you take a very small amount - like, way smaller than even a fair use - the courts don't care. If you're taking a handful of word probabilities from every book ever written, then the portion taken from each work is very, very low, so courts aren't likely to care. If you're only training on a handful of works then you're taking more from them, meaning…

If your training process ingests the entire text of the book, and trains with a large context size, you're getting more than just "a handful of word probabilities" from that book.

If you've trained a 16-bit ten billion parameter model on ten trillion tokens, then the mean training token changes 2/125 of a bit, and a 60k word novel (~75k tokens) contributes 1200 bits.

It's up to you if that counts as "a handful" or not.

Post reply on HN