Google Paid $2.7 Billion To Bring Back an AI Genius Who Quit in Frustration (msn.com) 47
At a time when tech companies are paying eye-popping sums to hire the best minds in artificial intelligence, Google's deal to rehire Noam Shazeer has left others in the dust. From a report: A co-author of a seminal research paper that kicked off the AI boom, Shazeer quit Google in 2021 to start his own company after the search giant refused to release a chatbot he developed. When that startup, Character.AI, began to flounder, his old employer swooped in.
Google wrote Character a check for around $2.7 billion, according to people with knowledge of the deal. The official reason for the payment was to license Character's technology. But the deal included another component: Shazeer agreed to work for Google again. Within Google, Shazeeer's return is widely viewed as the primary reason the company agreed to pay the multibillion-dollar licensing fee. The arrangement has thrust him into the middle of a debate in Silicon Valley about whether tech giants are overspending in the race to develop cutting-edge AI, which some believe will define the future of computing.
Google wrote Character a check for around $2.7 billion, according to people with knowledge of the deal. The official reason for the payment was to license Character's technology. But the deal included another component: Shazeer agreed to work for Google again. Within Google, Shazeeer's return is widely viewed as the primary reason the company agreed to pay the multibillion-dollar licensing fee. The arrangement has thrust him into the middle of a debate in Silicon Valley about whether tech giants are overspending in the race to develop cutting-edge AI, which some believe will define the future of computing.
I'm not an AI genius (Score:1)
But I would still like billions of dollars. Ideas?
Re:I'm not an AI genius (Score:5, Funny)
Create a startul called NotOnlyFans - this way, you branch out. Fans are nice, but they are just only one component of PCs.
Re:I'm not an AI genius (Score:4, Funny)
*startup, even.
Re: (Score:2)
Create a startup called NotOnlyFans - this way, you branch out. Fans are nice, but they are just only one component of PCs.
Better yet, create a site about what they're fans of, e.g. OnlyCPUs, OnlyHeatSinks, OnlyPSUs, etc.
Re: (Score:2)
Still too limiting...
Re: (Score:2)
OnlyThermodynamicTransfers.com
Re: (Score:2)
Ya don't say? (Score:5, Funny)
The arrangement has thrust him into the middle of a debate in Silicon Valley about whether tech giants are overspending in the race to develop cutting-edge AI, which some believe will define the future of computing.
There's a debate about it? I suppose Silicon Valley has always been a little fantasy based, though I think the AI hype cycle has been one of the bigger and flashier ones we've seen in a minute or two.
What we're seeing in the AI big boys club is the sunk cost fallacy on a scale rarely, if ever, seen before. Combined with the hype machine, it creates a absolute certainty that the end-result *HAS* to be all the amazing things folks like Sam Altmant say, or people wouldn't be spending all this money on it. And if you're going to play the game, you have to be the winner. He who spends the most is absolutely assured to win! AND THEN THEY CAN BE HUMANITY'S SAVIOR AND BE WORSHIPPED AND PRAISED FOREVERMORE!
That's the mentality of the AI companies in a nutshell. They believe all the hype, and absolutely believe that they are on the right path and that all they need is just more, more, MORE, MORE of everything to get there. More horsepower, more money, more energy, more datacenters, more. Greed on a scale the pre-digital robber barons would look at with a raised eyebrow and go, "Bro, like, could you maybe chill with the gimme gimme? You're embarrassing us."
Re: Ya don't say? (Score:2)
Re: (Score:3)
I'd say find people who "pragmatically understand this stuff". Guys like Noam Shazeer are generally high on their own supply and either never see the issues with AI or willfully ignore them.
Re: (Score:3)
Believe it or not, there's no shortage of people like that. What's hard to find are competent people who are also good at being dishonest grifters.
The con is getting really silly. Just look at the marketing for o1: "This one can actually reason!... er ... Forget about all the times we said the older models could do that. Breakthrough!" Now, as an honest person you might think that there was some fundamental change that affords new capabilities, but that isn't the case at all. There's nothing new here.
Re: (Score:2)
Everyone sees the potential for a second Industrial Revolution, or a new ubiquitous technology like Smartphones were.
They don't want to be caught letting the market slip them by.
Re: (Score:2)
Re:Ya don't say? (Score:4, Interesting)
SingularityNET: A decentralized marketplace for AI algorithms and services. It allows developers to monetize their AI solutions, and users can access a wide range of AI services. Transactions are facilitated using the platform's native cryptocurrency, AGIX.
Fetch.ai: Fetch.ai is building a decentralized platform that uses AI and machine learning to enable autonomous economic agents. These agents can perform tasks such as data analysis, supply chain optimization, and energy grid management. The platform operates using its native token, FET.
Ocean Protocol: This project provides a decentralized data exchange protocol that leverages blockchain technology to unlock data for AI. It enables secure and transparent data sharing, allowing AI models to access diverse datasets. The native token, OCEAN, is used for transactions within the network.
Numerai: Numerai is a hedge fund that crowdsources AI models from data scientists worldwide. Participants build predictive models using encrypted data and are rewarded in the platform's cryptocurrency, NMR, based on their model's performance in the stock market.
DeepBrain Chain: A decentralized AI computing platform that aims to reduce the cost of AI computations. By leveraging blockchain technology, it allows for secure and private sharing of computing resources. The network uses the DBC token for transactions.
Cortex: Cortex is developing a decentralized AI platform that integrates AI models into smart contracts. This allows for AI-powered decentralized applications (dApps) that can execute AI algorithms on the blockchain. The platform uses the CTXC token.
Re: (Score:2)
Those all sound like AI-generated taglines for companies founded by students in 6-month MBA programs.
Re: (Score:2)
Re: (Score:3)
For technical information, AI has largely replaced search for me. I'm not going to say this particular deal was worth what they paid. But google is not irrational to try very, very hard to maintain a stranglehold on the firehose of cash they currently have.
Re:Ya don't say? [You're spending too much...] (Score:3)
Your [nightflameauto's] comment is currently modded Funny, but maybe you can explain what they're laughing at? I think you're mostly poking at the pink elephant in the room that no one wants to think about. (Why don't they want to think about the room?)
I'm mostly thinking about it from the other perspective, which is that our perspective on AI must be all wrong. Enormous amounts of money and electricity and silicon are being poured into AI, but we keep going off the rails. AI winter follows AI summer and wh
Re: (Score:2)
It's a Mad Mad Mad Mad World. They all know (or believe) that there's a lot of money there at the destination, and they're going to crazy hijinks to try and get there first. Except that in the movie, hilarity ensued, whereas in real life it could cause an economic calamity similar to the dotcom or mortgage backed security bust.
Now if they paid those billions in stock, then when Google goes bust both the payer and payee lose out.
Re: (Score:2)
Buzzword bingo! (Score:0)
AI Genius
That's an oxymoron.
Literally, two fucking comments down from me.
AI Genius (Score:2)
The new Jumbo Shrimp.
Fuck all you trash.
AI Genius (Score:1)
A-B testing (Score:1)
Re: (Score:1)
it's like they just picked up random people at the bus stop and said "here, manage the biggest video platform in the world."
...that might actually produce positive results.
Monopolies & market consolidation (Score:5, Interesting)
This is why are great grand parents cranked tax rates on the top end of corporate earnings, banned stock buy backs and broke up the trusts. They'd be disappointed as heck with us if they saw us just sitting here taking it on the chin.
Re: (Score:2)
Sort of kind of (Score:2, Insightful)
We've been allowing market consolidation for decades and making companies pinky swear they wouldn't raise prices in order to continue the consolidation trend. It was always going to bite us in the ass hard it was a question of when not if.
COVID bit in the corporate profits massively as you might imagine and all that mo
Re: (Score:2)
Nominal profit increasing is not a sign of monopoly power. What you want to look at is increases in profit margin.
Profit margin (return on investment) is what goes up when you abuse market power, not merely raw profit numbers. I haven't looked to see the trends in profit margin in detail. Also note that abuse isn't the only way to improve margin, so it's not even as clear-cut as looking at that number.
You have to be diligent when looking at the numbers, I agree; because "profit" can look smaller if it's r
Re: (Score:2)
Why would Donald Trump do such a thing?
Re: (Score:2)
this is why inflation spiked
Ah, no. Inflation spiked for ordinary macroeconomic reasons having to do with supply chain crunches reducing supply while tax cuts and stimulus payments boosted demand. When supply goes down, prices go up. When demand goes up, prices go up.
Every time a mega corp screws up instead of being forced to compete they just take a mountain of cash from their untaxed income and use it to buy up the competitor.
Assuming the summary is correct, Google didn't really buy a competitor, because the company they bought was failing, not competing.
The same pattern here can be found in our food supply, automotive industry, heck even the housing industry.
Got any actual examples of that happening?
Re: (Score:2)
Capital is better remunerated than labor. Way better. Given all the power capital has, it's not surprising. The power of labor used to reside in labor organizations. Most workers are not considered labor though, because they're "white collar" and may get a limited amount of worthless RSU as part of their compensation.
Re: (Score:2)
this is why inflation spiked
Exactly! Just like how Led Zeppelin caused inflation in the 1970s. It's undeniable!
(Don't believe any slightly more logical associations, folks!)
Google PHB-to-English translation: (Score:3)
..."We're so desperate to win the AI market-share war that we decided to put up with Sheldon Neckbeard Cooper's many quirks. We'll kick his annoying ass out AFTER we win."
name sounds familiar. (Score:1)
Looking forward to the end (Score:3, Interesting)
AI is magic to consumers (and stops being magic and starts being something else the moment you explain it, exactly like magic).
AI's aesthetic is this century's Italian Futurism to weird techbros and the far right.
The faster this shit crashes like the value of Beanie Babies, the better off everyone else is.
Re: (Score:1)
Genuinely interested, what does the value of a Beanie Baby have to do with the welfare of those who are not interested in Beanie Babies?
Re: (Score:3, Insightful)
AI is magic to consumers (and stops being magic and starts being something else the moment you explain it, exactly like magic).
AI's aesthetic is this century's Italian Futurism to weird techbros and the far right.
The faster this shit crashes like the value of Beanie Babies, the better off everyone else is.
Your first sentence is correct.
I've no opinion on Italian Futurism.
Your third sentence is wrong. Even if it never gets any better it's going to severely alter a large number of jobs. This will be to the advantage of at least some people.
Too much? (Score:2)
Re: (Score:2)
>Obviously, nothing actually close to self-awareness level AI is going to happen until they spend $trillions on even bigger nucular-powered, quantum data-farms.
AI can be reached within a few years using less than 20W of power. We know because newborns are pretty damn intelligent in actually useful ways after just a few years.
Rather than working on the algorithms, we ought to be spending our money on the hardware. Emulating analog processes with binary switches is very wasteful.
Re: (Score:2)