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

361–370 of 466 posts

Re: Microsoft lays off one of its responsible AI teams

#361

Post by banannaise »

bigtunacan wrote: post #36
hn_throwaway_99 wrote: post #9

These sorts of adjunct, off-to-the-side "ethics" teams are always corporate window dressing, and when push comes to shove they always get jettisoned when competition demands an increase in speed. I don't mean to start a flame war, but this is a reason why I think "Chief Diversity Officers" are such an ill-conceived notion. It's not that I don't think diversity in corporate environments is extremely valuable, but it's…

This is certainly true in as much as I've experienced it in companies. The CDO (Chief Diversity Officer) has very little power and influence. Often it seems this is something many companies hire for so they can do the following 1. Say, "Look, we're trying here. We hired a CDO." 2. Have someone to take the blame if/when nothing changes. "We hired a CDO and nothing has changed, it must be their fault." I've seen some v…

It's not terribly dissimilar from the Glass Cliff. You set your diverse hires up for failure, and then they fail, and then that failure gets pinned on them. It may not be malicious in nature, but it can definitely have negative effects on their career paths.

Re: Microsoft lays off one of its responsible AI teams

#362

Post by charles_f »

In my opinion these AI committees would make sense with true AI being developed. We're not there yet, LLMs being advanced nlp and probabilistic discussion simulation.

With LLMs ethics are not much different than what you follow with regular computing - don't still people's data, don't give the people's data you stole anyway to other people, don't let the people you gave that stolen data do evil shit with it, apologize for the shit that people did with that stolen data you sold them and don't change anything.

ps: I'm a layman in llm so my opinion counts for shit

Re: Microsoft lays off one of its responsible AI teams

#363

Post by ben_w »

polygamous_bat wrote: post #308
ben_w wrote: post #220

Earlier quoted context omitted.

Sam Bankman-Fried is a uniquely terrible example of ideology, given St. Petersburg: https://conversationswithtyler.com/episodes/sam-bankman-frie... and https://astralcodexten.substack.com/p/open-thread-250 (and, y'know, being a fraudster who temporarily pulled the wool over the eyes of the regulators not just the charities he promised money to) Also: is SBF really the central example of alignment in your head, and no…

> Also: is SBF really the central example of alignment in your head, and not, say Yudkowsky? Because Yudkowsky is my central example, and he's the exact opposite on all counts. For normal people who don't spend six hours every day on lesswrong, they were literally in the same camp up until about six months ago (when all the "rationalists" violently disclaimed SBF after being funded by him for a few good years.)

Bearing in mind that I don't spend 6 hours a day on lesswrong (or even per month, not even if I include podcasts tangentially inspired by it), and I hadn't even heard of SBF until his crypto-thingie did what I expect every crypto-thingie to do, that caricature you're painting is going to be one of those ones that need a lot of labels on everything.

Re: Microsoft lays off one of its responsible AI teams

#364

Post by ausbah »

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…

I read your take as "ethics doesn't provide immediate economic value or model improvement along traditional metrics of performance, so companies are right to can them if they choose". I see ethics as looking at problem before they happens which is sort of an economic externality that companies won't pay for since the cost isn't something they have to bare, but in a more concerned world would.

Re: Microsoft lays off one of its responsible AI teams

#366

Post by flangola7 »

anonylizard wrote: post #19
hn_throwaway_99 wrote: post #9

These sorts of adjunct, off-to-the-side "ethics" teams are always corporate window dressing, and when push comes to shove they always get jettisoned when competition demands an increase in speed. I don't mean to start a flame war, but this is a reason why I think "Chief Diversity Officers" are such an ill-conceived notion. It's not that I don't think diversity in corporate environments is extremely valuable, but it's…

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…

I don't think there are any true corporate alignment teams. Anyone that actually believes in AI alignment would be slamming the brakes on AI development of any kind, lest the entire human population fall over dead suddenly.

Re: Microsoft lays off one of its responsible AI teams

#367

Post by belorn »

XorNot wrote: post #260
belorn wrote: post #214

Earlier quoted context omitted.

We have been here a long time with both fingerprints and DNA. A black box points the finger and it just happens to align with cultural bias and expectations. No one really want to change the system if there is a risk that a person people think is guilty might go free.

DNA testing is hardly a black box. The methodology, process, scientific basis and all other details are open and widely available. DNA has also cleared a lot of people who have been falsely accused.

People have ended up in jail with a death sentence based on DNA from a dog.

DNA in theory is very simple. It is much harder when you got a mix of multiple unknown donors, contaminated or decayed samples. Problems quickly occur when they then apply some "computer algorithms" to puzzle together things to create a probabilistic match. It is those proprietary black box computer algorithms that are being heavily discussed in courts, not the theory of DNA.

If we removed all cases that involved multiple unknown donors, contaminated or decayed samples, then DNA evidence would be much safer and open and we wouldn't need those proprietary computer algorithms. It would also significant reduce the number of cases where DNA evidence is used.

Re: Microsoft lays off one of its responsible AI teams

#368

Post by RandomLensman »

CuriouslyC wrote: post #355
RandomLensman wrote: post #317

Earlier quoted context omitted.

I think it is fair to say that "not at all possible" is perhaps too strong. On the potential of certain approaches I am less certain we have or know what we need. But we also have examples of things all around us we cannot make ourselves (e.g., a simple animal or lifeform from scratch).

Except that we've already created synthetic genomes: https://www.nytimes.com/2019/05/15/science/synthetic-genome-... If we can define what "generally intelligent" means in concrete terms, we can absolutely create it. The problem is that we don't have a good understanding/shared idea of what we're trying to achieve. We can create models that can learn to play many different games using the same weights, solve IQ/logic…

I don't share the view that all that is holding us back from creating AGI is a definition. Btw., that synthetic genome is a long way from making even something like an little nematode.

I rather think AGI "idealists" (maybe there is a better word) will forever hold out (any) algorithmic advances as a sign of the immanent creation of AGI proper.

Re: Microsoft lays off one of its responsible AI teams

#369

Post by flangola7 »

RandomLensman wrote: post #263
Jcowell wrote: post #253

Earlier quoted context omitted.

This seems dumb to me. It’s like if the Japanese had the clairvoyant ability to know that nuclear bombs were being built and used and the immediate future and said “We can ignore it until it exists)”. No. You leave Hiroshima and Nagasaki now and not when the freaking bombs are above your head.

We don't have the clairvoyant ability to know that AGI will exist in any future near enough to worry about. It might or it might not - and even if it exists we could get its nature wrong and worry about the wrong things.

This isn't congruent with the AI research community consensus. There's debate over many points, but that we will develop AGI is seen as a forgone conclusion almost universally. The most pessimistic place it closer to 2100, others say 5 years, most say Notably the people closest to the fire, those advancing SOTA AI capabilities and those deepest into alignment/existential risk research, both say it is coming this decade. They predict very different outcomes (crudely summarized: utopia vs extinction), but they both predict the event horizon is on our doorstep.

You're right that we might get it wrong, but getting things wrong with a new technology really always means negative outcomes. When you're figuring out rocket science and you realize you worried about the wrong things your rocket doesn't somehow still succeed and make it to space, it blows up in a fireball. It might not blow up because of the specific thing you were paying attention to, but it still blew up.

Re: Microsoft lays off one of its responsible AI teams

#370

Post by sebzim4500 »

xpe wrote: post #337
sebzim4500 wrote: post #316

Earlier quoted context omitted.

A sufficiently good LLM would not produce neonazi propaganda when asked to write a HS paper, regardless of whether it had been 'aligned' or not.

Why not? Can you unpack what you mean? Are you saying that sufficiently good models should understand that such propaganda is not appropriate for the context? Are you saying that understanding appropriateness is not the same thing as ethics?

I mean that the model should learn that e.g. highschool textbooks are not filled with neonazi propaganda, therefore you should not produce it when asked to write a highschool essay. I would assume that if you go out of your way you can make LLaMA generate such content, but probably not if you fill the prompt with normal looking schoolwork.

This is completely orthogonal to learning what is ethically right or wrong, or even what is true or false.

Post reply on HN