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

Meta invests $14.3B in Scale AI to kick-start superintelligence lab

nytimes.com

Post reply on HN
Print view View original

231–240 of 507 posts

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#231

Post by ryanblakeley »

Bjorkbat wrote: post #216
rybosome wrote: post #124

Earlier quoted context omitted.

Scale has also built massive amounts of proprietary datasets that they license to the big players in training. Meta, Google, OpenAI, Anthropic, etc. all use Scale data in training. So, the play I’m guessing is to shut that tap off for everyone else now, and double down on using Scale to generate more proprietary datasets.

It's a smart purchase, it's just that I don't see how these datasets factor into super-intelligence. I don't think you can create a super-intelligent AI with more human data, even if it's high-quality data from paid human contributors. Unless we watered-down the definition of super-intelligent AI. To me, super-intelligence means an AI that has an intelligence that dwarfs anything theoretically possible from a human m…

Super-intelligent game-playing AIs, for decades, were trained on human data.

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#233

Post by sensanaty »

epsilonthree wrote: post #224

I work at Meta. Scale has given us atrocious data so many times, and gotten caught sending us ChatGPT generated data multiple times. Even the WSJ knew: https://www.wsj.com/tech/ai/alexandr-wang-scale-ai-d7c6efd7 https://archive.is/5mbdH We intentionally didn’t use them at all for Llama 2 and mostly avoided using them for Llama 3, but execs kept pushing Scale on us. Total mystery why until now, guess this explains it.

Interesting that your comment got flagged almost instantaneously

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#234

Post by MichaelZuo »

tyre wrote: post #157
DebtDeflation wrote: post #151

Earlier quoted context omitted.

It's a wild story for sure. Dropped out of MIT after freshman year and starts Scale to do data labeling. Three years later Scale has a $1B valuation and two years after that Wang is the world's youngest billionaire. Nine years after Scale's founding they're still doing less than $1B in annual revenue. Yet Meta is doing a $14B acquihire. There's definitely more than meets the eye. I suspect it involves multiple world…

> Dropped out of MIT after freshman year and starts Scale to do data labeling I was in their YC batch, so two notes: 1. He didn't start it himself 2. They weren't doing data labeling when they entered YC. They pivoted to this.

There are some stranger rumors too floating around.

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#235

Post by HardCodedBias »

A few possibilities:

1. Mark no longer wants to run the company and he is picking alexander wang. 2. Mark believes that Ai is the top priority, his teams have failed, (this is all clearly true so far) and he wants completely change the org structure of his AI efforts (not recommendation but everything else). 3. Mark wants to cut off the supply of information to other labs 4. Mark thinks that full access to ScaleAi's data could accelerate their research and somehow they couldn't do this with a less expensive options.

(2) seems semi-reasonable (in that Meta has failed with near infinite resources) but acquiring a handful of execs for this price seems absurd.

(3) seems like a conspiracy theory and the technology is moving away from this path of data collection, although it is still important at this very moment.

(4) Maybe.

I guess some combination of all 4 is plausible. But the amount of money seems, frankly, absurd.

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#237

Post by paxys »

noosphr wrote: post #223
paxys wrote: post #178

Earlier quoted context omitted.

People on HN aren't the ones driving Ubers, so I'm not sure what experiences you are expecting to hear about. Go talk to actual drivers and you'll find that things aren't exactly rosy.

Driving for Uber in 2010 was amazing. There were plenty of people on HN who signed up for the app to drive people back home before and after work. Being able to see your car move in real time on the uber database with >2s lag between your car GPS and customers phone was magical in a way that's hard to describe today.

Lots of things were amazing when there was unlimited VC money flowing in and no expectation of profit. No point bringing it up 15 years later in a new reality.

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#238

Post by Fraterkes »

Bjorkbat wrote: post #216
rybosome wrote: post #124

Earlier quoted context omitted.

Scale has also built massive amounts of proprietary datasets that they license to the big players in training. Meta, Google, OpenAI, Anthropic, etc. all use Scale data in training. So, the play I’m guessing is to shut that tap off for everyone else now, and double down on using Scale to generate more proprietary datasets.

It's a smart purchase, it's just that I don't see how these datasets factor into super-intelligence. I don't think you can create a super-intelligent AI with more human data, even if it's high-quality data from paid human contributors. Unless we watered-down the definition of super-intelligent AI. To me, super-intelligence means an AI that has an intelligence that dwarfs anything theoretically possible from a human m…

I think this is kind of a philosphical distinction to a lot of people: the assumption is that a computer that can reason like a smart person but still runs at the speed of a computer would appear superintelligent to us. Speed is already the way we distinguish supercomputers from normal ones.

Re: Meta invests $14.3B in Scale AI to kick-start superintelligence lab

#240

Post by OtherShrezzing »

vonneumannstan wrote: post #114
akudha wrote: post #84

Earlier quoted context omitted.

What is the big risk here? He has the cash to burn and he has full control of his position. Nothing will happen to him if he wastes a few Billions. He is not some poor single mom who has to decide between fixing her car and paying for kid’s Christmas presents

>What is the big risk here? That this is clearly the wrong person to hire? Maybe Demis or Ilya is worth $15B but Wang? Extremely odd choice...

I don’t think either Demis or Ilya are available for $15bn. They’re already both comfortably billionaires. Demis seems a sensible candidate for heir to the top job at Google in the long term. Ilya is away focusing on super intelligence.

It’s not clear to me why either would take a subservient role in a company flailing incoherently around AI, rather than stick with the incredibly high-leverage opportunities they both have now.

Post reply on HN