AI: DeepMind and Brain team up to compete for market – 04/28/2023 – Tech

Google founder Larry Page convinced Demis Hassabis to sell his artificial intelligence company DeepMind with a pledge. The London-based startup would be under no pressure to make money in order to focus on a single goal: creating computer software that equals or surpasses human intelligence.

Since the £400m deal in 2014, Hassabis has struggled to keep Page’s promise and, according to three people briefed on the efforts, has gone even further. DeepMind pushed for an independent legal status, similar to a not-for-profit organization, with an independent governance board overseeing the powerful technology it was trying to build.

Such moves came to an end last week, however, when Sundar Pichai, Page’s successor as head of Google’s parent company Alphabet, announced that DeepMind would merge with Google Brain, the California-based tech giant’s AI lab. .

The move means Hassabis will cede DeepMind’s cherished independence in exchange for greater power and influence over the future of AI. The newly formed “Google DeepMind” unit will be led by him, with the clear mission to develop “general AI systems” even more “capable and responsible” and that can be integrated into new products and services, according to Pichai.

Google’s reorganization was triggered by the rise of OpenAI, the Microsoft-backed group that last November launched ChatGPT, a chatbot that provides plausible, nuanced text responses to questions.

The sudden popularity of ChatGPT, used by more than 100 million people in January, shattered Google’s belief that it had a strong lead in the race to build and commercialize AI.

Microsoft has since announced that it will incorporate tools created by OpenAI into its productivity software. Wealthy rivals from Meta to Alibaba have announced big investments in AI products. Several well-resourced startups, including a nascent effort by Tesla and Twitter boss Elon Musk, are expected to offer even more competition.

Google also faces an existential dilemma. Its search business, which accounted for most of Alphabet’s $283 billion in annual revenue in 2022, appears to be under direct threat. Large language models like ChatGPT can respond to queries with comprehensive written answers rather than a set of links. These tools can provide millions of users with a route around online advertising.

Seeking to respond to the threat, Google has reorganized itself so that Hassabis is more clearly in charge of the company’s operations, according to people with knowledge of the measures.

To accelerate the search giant’s efforts, Hassabis said his new team “would need to work with greater speed, stronger collaboration and execution, and streamline the way we make decisions to focus on achieving the greatest impact.”

To take on OpenAI, the newly formed Google DeepMind unit will need to put years of rivalry aside, according to several former Google and DeepMind employees and their collaborators.

While Google Brain was also made up of elite AI researchers, the two teams had distinct cultures. London-based DeepMind was considered the crown jewel of the company, largely isolated from the rest of the group and taking a top-down approach, according to four former employees of both organizations.

He often worked on classified projects, although that became more difficult as they began to consume more and more computing power in Google’s data centers. His mission to “solve” intelligence was sacred. The success of its employees was measured through the publication of papers in top scientific journals, such as Nature. Three people close to DeepMind said its leaders are focused on “Nobel-level problems”, meaning that if solved they would be worthy of a Nobel Prize.

Meanwhile, California-based Brain was more open and unstructured. It has contributed significantly to Google’s bottom line over the years, according to several former employees. Brain production has been integrated into search, advertising, translation and other parts of Google’s business.

Despite these successes, employees say they began to feel adrift. “I felt that the organizational setup [na Brain] it was really, from the beginning, something that couldn’t scale,” said a former Google employee.

Brain’s leader, Jeff Dean, an experienced and highly respected AI scientist at the company, was reluctant to make the tough decisions needed to focus and scale ideas, such as addressing staffing issues or weeding out flawed projects, the source said.

“If I had to sum up why OpenAI is where it is with so many fewer resources and people compared to Google, it’s because they have provided a clear mission. The brain has lost its way,” he added.

According to Pichai, last week’s revamp would allow Dean to focus on research, leaving Hassabis in charge of an expanded group of scientists.

While the changes are likely to “ruffle a lot of feathers” and cause some staff turnover, this could be Google’s best chance to run “a stiffer ship” and catch up with its competitors, according to people briefed on the merger.

The two AI teams are already responsible for major advances in artificial intelligence. DeepMind’s contributions include AlphaFold software, which can predict the shape of nearly all known proteins. Brain’s include Transformer, the AI ​​technique that is the foundation of the new wave of generative AI products, including ChatGPT and Google’s own Bard chatbot.

Over the past two years, Hassabis has come to terms with the need for greater collaboration with Google, according to people close to the company.

The amount of computing power and investment required to build so-called large language models, which can generate sophisticated and plausible text, has expanded rapidly. Recruiting AI talent is increasingly expensive. Microsoft announced this year that it would invest up to $10 billion in OpenAI over several years to address these challenges.

The rise of Hassabis also means that the center of Google’s AI operation moves to London, which will have profound implications.

“How much influence DeepMind holds (…) greatly affects how accountable [o Google] will be,” said Dan Hendrycks, director of research at the Center for AI Safety [Centro para Segurança de IA]. He noted that Shane Legg, co-founder of DeepMind, has been focused on security and AI risk for over a decade, while Google Brain lacked a dedicated security or alignment team. Google said it had hundreds of people working on responsible AI across the company.

People close to the new division said they believed the merger would be beneficial for AI research in general. Pichai said the most “critical and strategic” AI design would involve building a series of “powerful multimodal AI models”. These would be similar to GPT-4, which powers part of Microsoft’s Bing search engine, and LamDA, the Google model that powers its Bard chatbot.

But it was not yet clear how the reorganization would result in more ambitious commercial products, the people said. After all, DeepMind has never actively advocated the need to earn money from its work. People close to Hassabis argue that he understands that this is the time to change.

“Demis is very pragmatic and highly competitive,” said a former DeepMind employee. “He’s ideological in terms of … vision, but he wants to be on the edge of that. He’s not anti-commercial. That’s the best way for him to do the job, or he wouldn’t have taken it.”