Home About us

Author of Transformer's "biggest contribution", he returned to Google as co-head of technology at Gemini

Qubits 2024/08/23 12:24

Co-Head of Technology at Gemini!

That's the latest move from Noam Shazeer, the author ·of Transformer's "biggest contribution", since his return to Google.

Qubits, the author of Transformer's "biggest contribution", returned to Google as co-technical director of Gemini

According to more leaks from The Information, Shazeer will work on Gemini along with Jeff Dean, Google's head of AI, and Oriol Vinyals, chief scientist at DeepMind.

As a project that directly benchmarks against OpenAI's GPT model, Gemini can be said to be becoming more and more lively with the return of the original author of Transformer.

"Packaged and recycled" by Google

In fact, before this revelation, Google had already made a "trenched" acquisition at the beginning of this month-

Pack and take away the core team of the star AI unicorn Character.AI.

Among them are the two founders, Noam Shazeer and Daniel De Freitas.

Qubits, the author of Transformer's "biggest contribution", returned to Google as co-technical director of Gemini

Google's price at the time was $2.5 billion (about 17.9 billion yuan), much higher than Character.AI $1 billion valuation.

Although most of the Character.AI team stayed to continue building Character.AI products, netizens still discovered Huadian:

Isn't that just a shell left?

Qubits, the author of Transformer's "biggest contribution", returned to Google as co-technical director of Gemini

Shazeer's relationship with Google goes back even further.

In 2021, Shazeer left Google and founded Character.AI with colleague Freitas.

While at Google, they were mainly responsible for building LaMDA, a language model for AI conversations.

Shazeer is one of the eight authors of Transformer, and is also recognized as the "biggest contributer" -

Based on his own ideas, he rewrote the entire project code, taking the system to the next level, making the Transformer project "the prelude to the sprint".

The main reason for founding Character.AI is to further research on more personalized superintelligence.

No, after more than two years of development, Character.AI has accumulated more than 20 million users with "various AI roles".

In March 2023, Character.AI closed a $150 million funding round at a $1 billion valuation, led by a16z, with participation from former GitHub CEO Nat Friedman, Elad Gil, A Capital, and SV Angel.

But after that, the star AI unicorn began to get into trouble:

Four months after this financing, news broke that a new round of financing negotiations was underway Character.AI, but there was no follow-up since then.

In the past year, Character.AI has frequently reported news of poor financing. On July 4, Character.AI was revealed to be considering selling to Google and Meta.

And now, the result of Character.AI is also sold to Google in disguise, after all, even the core team has been taken away.

AIGC's "Common Problem"

If you think about it carefully Character.AI behind the results, there are actually some "common problems" of AIGC startups——

That is burning money and lack of money.

It is not difficult to understand how to burn money, after all, the expensive computing power required to train large models is a headache for many enterprises, not to mention the fierce competition, and all players are in full swing to update and iterate.

The lack of money means that the AIGC product business is very difficult to monetize.

and Character.AI staged the "same operation", as well as Inflection, Adept, etc.

Microsoft paid to license Inflection's technology, and the founder of Inflection joined Microsoft directly with most of its employees to form a new division.

Similarly, Adept, a startup founded by Amazon and another Transformer author, has done something similar and has been investigated by the United States Federal Trade Commission (FTC) for avoiding antitrust investigations through technology licensing.

In addition, there is also Stability AI, which is the founder and core team, and Perplexity, the AI search leader...... There are rumors circulating that they are seeking acquisitions.

And the latest news is that Perplexity announced:

The company plans to start advertising on the search app in Q4.

Perplexity has adapted the way it presents information sources and introduced a revenue sharing model that allows publishers to earn revenue through search engines.

Its advertising strategy will adopt a CPM (cost per thousand impressions) model, with a price of more than US$50 (about 357 yuan) according to people familiar with the matter, and the initial ad categories will include technology, health and pharmaceuticals.

This also makes people have to sigh that even AI search has to rely on advertising in terms of profit model, just like traditional search.

This article is from Xinzhi self-media and does not represent the views and positions of Business Xinzhi.If there is any suspicion of infringement, please contact the administrator of the Business News Platform.Contact: system@shangyexinzhi.com