OpenAI departures: Why can’t former employees talk? • Hacker News Board

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

#341

Post by resonious » Sat, May 18, 2024, 5:05 AM UTC

brudgers wrote: post #287

fshbbdssbbgdd wrote: post #227

Earlier quoted context omitted.

In the past a lot of options would expire if you didn’t exercise them within eg. 90 days of leaving. And exercising could be really expensive. Speculation: maybe the options they earn when they work there have some provision like this. In return for the NDA the options get extended.

My unreliable memory is Altman was ( once? ) in favor of extending the period for exercising options. I could be wrong of course but it is consistent with my impression that making other people rich is among his motivations. Not the only one of course. But again I could be wrong.

Wouldn't be too surprised if he changed his mind since then. He is in a very different position now!

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

#342

Post by d0mine » Sat, May 18, 2024, 5:06 AM UTC

YurgenJurgensen wrote: post #45

a_wild_dandan wrote: post #16

Is this a legally enforceable suppression of free speech? If so, are there ways to be open about OpenAI, without triggering punitive action?

I believe a better solution to this would be to spread the following sentiment: "Since it's already illegal to tell disparaging lies, the mere existence of such a clause implies some disparaging truths to which the party is aware." Always assuming the worst around hidden information provides a strong incentive to be transparent.

I hope forbidding telling the truth is about something banal like "fake it until you make it" in some of OpenAI demos. The technology looks like magic but plausible to implement in a few months/years.

Worse if it is related to training future super intelligence to kill people. Killer drones are possible even with today's technology without AGI.

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

#343

Post by TeMPOraL » Sat, May 18, 2024, 5:06 AM UTC

bluefirebrand wrote: post #339

TeMPOraL wrote: post #332

Earlier quoted context omitted.

Clever, but no. The argument about LLMs not being copyright laundromats making sense hinges the scale and non-specificity of training. There's a difference between "LLM reproduced this piece of copyrighted work because it memorized it from being fed literally half the internet ", vs. "LLM was intentionally trained to specifically reproduce variants of this particular work". Whatever one's stances on the former case,…

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

You may or may not agree with it, but that's the only thing that makes it different - scale and non-specificity. Same thing that worked for search engines, for example.

My point isn't to argue merits of that case, it's just to point out that OP's joke is like a stereotypical output of an LLM: seems to make sense, but really doesn't.

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

#344

Post by NewJazz » Sat, May 18, 2024, 5:07 AM UTC

bluefirebrand wrote: post #339

TeMPOraL wrote: post #332

Earlier quoted context omitted.

Clever, but no. The argument about LLMs not being copyright laundromats making sense hinges the scale and non-specificity of training. There's a difference between "LLM reproduced this piece of copyrighted work because it memorized it from being fed literally half the internet ", vs. "LLM was intentionally trained to specifically reproduce variants of this particular work". Whatever one's stances on the former case,…

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.

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

#345

Post by romwell » Sat, May 18, 2024, 5:08 AM UTC

mvdtnz wrote: post #319

crooked-v wrote: post #312

Earlier quoted context omitted.

Agreements with surprise terms that only get detailed later tend not to be very legal.

How do you know there isn't a very clear term in the employment agreement stating that upon termination you'll be asked to sign an NDA on these terms?

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.

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

#346

Post by wruza » Sat, May 18, 2024, 5:12 AM UTC

ihumanable wrote: post #130

ndriscoll wrote: post #116

Earlier quoted context omitted.

That doesn't really contradict what the other poster said. They're calling for regulation (digging a moat) to ensure systems are "safe" and "aligned" while ignoring that humans are not aligned, so these systems obviously cannot be aligned with humans; they can only be aligned with their owners (i.e. them, not you).

Alignment in the realm of AGI is not about getting everyone to agree. It's about whether or not the AGI is aligned to the goal you've given it. The paperclip AGI example is often used, you tell the AGI "Optimize the production of paperclips" and the AGI started blending people to extract iron from their blood to produce more paperclips. Humans are used to ordering around other humans who would bring common sense and…

AGI started blending people to extract iron from their blood to produce more paperclips

That’s neither efficient nor optimized, just a bogeyman for “doesn’t work”.

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

#347

Post by romwell » Sat, May 18, 2024, 5:14 AM UTC

jbernsteiniv wrote: post #204

jay-barronville wrote: post #173

It probably would be better to switch the link from the X post to the Vox article [0]. From the article: “““ It turns out there’s a very clear reason for [why no one who had once worked at OpenAI was talking]. I have seen the extremely restrictive off-boarding agreement that contains nondisclosure and non-disparagement provisions former OpenAI employees are subject to. It forbids them, for the rest of their lives, fr…

He gets my respect for that one both publicly acknowledging why he was leaving and their pantomime. I don't know how much the equity would be for each employee (the article suggests millions but that may skew by role) and I don't know if I would just be like the rest by keeping my lips tight for fear of the equity forfeiture. It takes a man of real principle to stand up against that and tell them to keep their money…

>It takes a man of real principle to stand up against that and tell them to keep their money if they can't speak ill of a potentially toxic work environment.

Incidentally, that's what Grigory Perelman, the mathematician that rejected the Fields Medal and the $1M prize that came with it, did.

It wasn't a matter of an NDA either; it was a move to make his message heard (TL;DR: "publish or perish" rat race that the academia has become is antithetical to good science).

He was (and still is) widely misunderstood in that move, but I hope people would see it more clearly now.

The enshittification processes of academic and corporate structures are not entirely dissimilar, after all, as money is at the core of corrupting either.

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

#348

Post by otterley » Sat, May 18, 2024, 5:16 AM UTC

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…

IAAL (but not your lawyer and this is not legal advice).

That’s not how it works. It doesn’t matter if you write the words yourself or have an agent write them for you. In either case, it’s the communication of the covered information that is proscribed by these kinds of agreements.

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

#349

Post by oblio » Sat, May 18, 2024, 5:16 AM UTC

saalweachter wrote: post #253

arthurcolle wrote: post #247

Earlier quoted context omitted.

No company? Are you sure? Aramco?

Apple has spent $650 billion on stock buybacks in the last decade. Granted, that might be most of the profit they have made, but still, they're probably at at least 0.7T$ so far. I bet they'll break $1T eventually.

Based on this they've had $1tn profits since 2009: https://companiesmarketcap.com/apple/earnings/

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

#350

Post by calibas » Sat, May 18, 2024, 5:16 AM UTC

jay-barronville wrote: post #173

It probably would be better to switch the link from the X post to the Vox article [0]. From the article: “““ It turns out there’s a very clear reason for [why no one who had once worked at OpenAI was talking]. I have seen the extremely restrictive off-boarding agreement that contains nondisclosure and non-disparagement provisions former OpenAI employees are subject to. It forbids them, for the rest of their lives, fr…

> It forbids them, for the rest of their lives, from criticizing their former employer.

This is the kind of thing a cult demands of its followers, or an authoritarian government demands of its citizens. I don't know why people would think it's okay for a business to demand this from its employees.