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

Microsoft lays off one of its responsible AI teams

platformer.news

Post reply on HN
Print view View original

181–190 of 466 posts

Re: Microsoft lays off one of its responsible AI teams

#181

Post by Tenoke »

matusp wrote: post #158
anonylizard wrote: post #19

Earlier quoted context omitted.

This. People should understand AI safety != AI ethics. AI safety has two branches, AI ethics and AI alignment. Those two branches despise each other with intense passion. The alignment branch (responsible for developing RLHF->ChatGPT) believes AI ethics is completely trivial, and distracts attention away from existential risks. The ethics branch (responsible for all the media attention about racist AIs/bias etc) view…

> The alignment branch (responsible for developing RLHF->ChatGPT) RLHF was not conceived by AI Alignment people. Using RL to train generative models was a thing even ten years ago. Now they finally made it work on scale. This has nothing to do with alignment.

A big motivation, and at minimum a solid chunk of the RLHF tuning does seem to have been alignment-focused.

Re: Microsoft lays off one of its responsible AI teams

#182

Post by hutzlibu »

csomar wrote: post #141
frutiger wrote: post #131

Earlier quoted context omitted.

> These were designed for over millions of years in differing circumstances. Designed or occurred by accident?

Designed by randomness?

Designed by rough selection for "does it work".

Re: Microsoft lays off one of its responsible AI teams

#183

Post by jeremyjh »

heretoo wrote: post #140
ketzo wrote: post #104

Earlier quoted context omitted.

Yeah, this is like claiming lead ethyl in gas “hasn’t caused mass harm” in 1954. There’s absolutely zero way to know that yet.

Except that lead's toxicity has been known for thousands of years.

Yes and the dangers of AI overlords have been known at least since the first Terminator movie /s.

Seriously though, people tend to dismiss things they are know are bad if there is no immediate harm. Certainly they had no idea how incredibly pervasive the lead from gasoline would be, that it would be detectable at harmful levels in basically everyone. It is also hard to measure these harms. How much crime of the 70s and 80s could be attributable for example.

Re: Microsoft lays off one of its responsible AI teams

#184

Post by narag »

skissane wrote: post #168
cameldrv wrote: post #79

Earlier quoted context omitted.

The main issue I see is terminology. Ethics is a broad field. “AI Ethics” has come to mean mostly eliminating different kinds of bias from AI models. However, AI raises multitudes of ethical questions, and merely being unbiased is far from enough for an AI model (or a person) to be acting ethically.

Whenever I hear the word “ethics” I always ask myself “whose”. I hate how people talk about ethics as if it was some non-controversial topic where everyone can clearly see what the “right” answers are. This world features a competition between radically opposed ethical systems. We don’t agree on whether ethics is objective (i.e. some people’s ethics are superior to those of others) or subjective (i.e. “superior” is j…

Often when they say ethics they mean politics, their politics specifically.

For me, democracy is a system where persons can vote for their interests. Different persons have different interests and that's good. You must compromise with others so everybody gets partly satisfied.

Others see democracy as a way to establish truth, those are the ones talking about ethics. Even if they lose, they still believe they must win, because they're right and the others wrong.

Re: Microsoft lays off one of its responsible AI teams

#185

Post by hutzlibu »

adamisom wrote: post #18
pixl97 wrote: post #17

Earlier quoted context omitted.

"If I don't make the world eating machine, someone else might make the world eating machine first!" No wonder we don't see aliens filling the universe. Intelligence is its own great filter.

Nah man, it’s just math! Whereas we have souls … or something.

We have consciousness. It is going to be hard to proof, that AI's can have that, too.

I mean, there were screenshots were the AI said it has it - qnd there will be more, but having something, is something different, from statistically inserting fitting strings.

Re: Microsoft lays off one of its responsible AI teams

#186

Post by Veen »

zirgs wrote: post #175
anonylizard wrote: post #19

Earlier quoted context omitted.

This. People should understand AI safety != AI ethics. AI safety has two branches, AI ethics and AI alignment. Those two branches despise each other with intense passion. The alignment branch (responsible for developing RLHF->ChatGPT) believes AI ethics is completely trivial, and distracts attention away from existential risks. The ethics branch (responsible for all the media attention about racist AIs/bias etc) view…

SD doesn't default to creating black people, but it understands the n-word. Should AI understand racial slurs?

To anthropomorphize, how would an AI know that a slur is a slur and therefore undesirable if it doesn't understand it. Little kids often accidentally say slurs they've heard in the media or playground until someone explains what it means and why it's bad.

Re: Microsoft lays off one of its responsible AI teams

#187

Post by IshKebab »

dmix wrote: post #115

Anyone paying close attention to the output of generic "AI ethics" people (especially on Twitter) should not be surprised that companies that when push comes to shove the businesses getting actually useful AI tools that are being tested by real people for real things generating real data, are starting to question the utility of the bulk of the crowd who skirted off hypotheticals for years to build their reputations a…

Yeah I think the question is are these people just philosophers pontificating about AI doom, or are they AI researchers creating techniques to make AI decisions less opaque and biased (or less biased in undesirable ways at least)?

The latter are definitely useful but the former is something you start so you can look good and then disband a few years later when they produce little useful output and cost a ton of money.

Re: Microsoft lays off one of its responsible AI teams

#188

Post by munchbunny »

daydream wrote: post #13

I’ve seen similar dynamics at other companies. When teams like this MS ethical AI group are separate organizationally from product and research and engineering teams, it’s harder to have an impact. Management and the product teams want to ship features. Research wants to work in cool new stuff. Nobody wants people from random other teams getting in their way and telling them to slow down. It’s hard enough getting mea…

Indeed, we’ve seen this movie before. If the CEO really wants some AI product to exist, an AI or even normal ethics team will end up getting steamrolled by any number of product teams in order to make an AI product happen sooner, actual ethics be damned. I’m actually really grateful that GDPR landed a few years before ChatGPT did because at least there’s a regulatory fear of god keeping product teams honest on the privacy aspect of AI ethics.

Re: Microsoft lays off one of its responsible AI teams

#189

Post by dragonwriter »

circuit10 wrote: post #179
dragonwriter wrote: post #163

Earlier quoted context omitted.

> AI alignment people are the group who are capable of bringing AGI to life, but want to do it responsibly. AI alignment people are the group who believe “responsibly” means, centrally, bringing AGI to existence as fast as possible, and, critically, under the tight, opaque, authoritarian control of a narrow group of people whose ideology is, well, exemplified by figures like Sam Bankman-Fried, and doing everything po…

> AI alignment people are the group who believe “responsibly” means, centrally, bringing AGI to existence as fast as possible From what I’ve seen it’s the exact opposite? They want AI development to slow down so we can get it right and make it safe. There are people who are trying to speed it up but that seems like the opposite of alignment to me

> From what I’ve seen it’s the exact opposite? They want AI development to slow down so we can get it right and make it safe.

The only way that the alignment faction wants to slow things down is in that they want to maintain tight control of models and gated access to assure “proper” use. This does have an effect of slowing certain kinds of development, as broader access to build around models without centralized controls fosters some aspects of development (the “Stable Diffusion moment”) but in ideal terms, the alignment faction is about fast-but-narrowly-controlled development.

The ethics faction is more about slowing things down – though more about adoption, particularly in sensitive uses, than development – but is more oriented toward openness/transparency.

Re: Microsoft lays off one of its responsible AI teams

#190

Post by IshKebab »

kmeisthax wrote: post #85
IceHegel wrote: post #28

You either do AI and talk about AI ethics (OpenAI) or do AI ethics and talk about AI (Google).

Google absolutely does a LOT of AI. And OpenAI stopped releasing model weights and switched to SaaS specifically citing ethical and safety concerns.

They said it was because of ethics. It was pretty transparently because of money.
Post reply on HN