The greatest personalities of artificial intelligence discussed the dangers of the rapidly progressive technology at the World Economic Forum this week when the hype around Donald Trump was created by a 500 billion dollar-Ki infrastructure project.
AI pioneers, including Sir Demis Hassabis, head of Google Deepmind, Dario Amodei, co-founder of Anthropic, and Yoshua Bengio, the “Godfather of the AI”, computer scientist, used the meeting in Davos to confirm significant warnings about AI threats,, Because commercial interests and geopolitical rivalries suppress security concerns.
Hassabis admitted that “the spirit cannot be put back into the bottle,” he said Artificial general intelligence – If computers exceed human cognitive skills – civilization could threaten if it gets out of control or is hijacked by bad actors. This is particularly the case with large voice models that are “open source” and accessible to everyone.
“This is much more than just a company or products,” said the Nobel Prize winner in an interview with the Financial Times. “(It is) the future of humanity, human constitution and where we want to go as a society.”
Amodei, whose start-up produces and is the chatbot Claude Supported by Google And Amazon said he was concerned about the use of AI by authoritarian governments and was “very concerned about the scenarios of 1984 or worse”.
“Science does not know how we can control machines that are only at our intelligence level, and even worse if they are smarter than we do,” added Bengio during a discussion round. “There are people who say: ‘Don’t worry, we’ll find it out.’ But do you understand the consequences if we don’t find out? ”
Her attitude was criticized by Yann Lecun, chief AI scientist at Meta, who spent billions for the development of an open source LLAM called Llama as hypocritical. He said that this concerns would be refuted by the hard competition of his competitors for the construction and sale of the best models.
“Yoshua and Dario commented against open source, and that is actually very dangerous,” he said in an interview. “Obstacles in the spread of open source would mean that a few players, either on the west coast of the USA or in China, would take over the regulation.” . Put power into the hands of a small number of people.
“It is very strange from people like Dario. We met yesterday, where he said that the advantages and risks of AI were roughly the same size, and I asked: ‘If you really believe that, why do you continue to work on AI?’ “So I think that he is a little ambiguous in this regard.”
While scientists and engineers debated KI about the risk-revenue ratio, managers from the economy showed unrestricted enthusiasm for the technology.
“There are no adversaries,” said Ervin TU, President of the Dutch Technology Investment Group Prosus. “If you have an understanding of what large voice models and agents trained on them can make it difficult as a person not to come to the conclusion that they are transformative and will be incredibly disruptive in every industry.”
On Wednesday the feverish atmosphere was further fueled by Openai, Softbank and Oracle If an American AI infrastructure-joint venture, announces $ 500 billion Called “Stargate”.
Trump received his managing directors Sam Altman, Masayoshi Son and Larry Ellison in the Oval Office on Tuesday before signing implementation regulations this week that would eliminate many obstacles to the development of the technology. The new US president said the steps would ensure American supremacy in technology.
“At Openaai, we believe that infrastructure is fate,” said Sarah Friar, CFO of Openaai. “Stargate is about more computing power. More computing power leads to better models. Better models solve more complex problems and offer more advantages for people and companies. “
For the rest of the week, Stargate dominated the debate in Davos, with many, including Elon Musk, questioned on his social networking page X how the trio was would finance the promised enormous expenses.
The FT reported on Friday that Stargate had not yet secured the necessary financing, no state financing would be received and would only serve Openai Completed once. Softbank and Openaai previously intended to provide more than $ 15 billion for the project, in the hope of raising a combination of equity from their existing donors and debts to finance Stargate.
The new company was also evaluated as the latest proof of a break in the relationship between Altman and Microsoft boss Satya Nadella and his top AI manager Mustafa Suleyman, the former co-founder of Deepmind, who left his own startup and at the beginning of last year to Microsoft changed.
“The tensions that came up in Davos between Mustafa Suleyman and Sam Altman last year were just the beginning,” said Salesforce boss Marc Benioff, who competed with Microsoft when selling AI-supported agents.
“Microsoft is now accelerating his own AI development. . . This pattern reflects the story of Microsoft with its ‘partners’, ”added Benioff. “This could mean the beginning of the end of the relationship and make it crucial for Openai to quickly expand to other platforms.”
“Marc has no idea what he is talking about,” said Microsoft spokesman Frank Shaw.
Microsoft Invested almost $ 14 billion in Openaai since 2019 and, in return, negotiated rights to his intellectual property and being his exclusive cloud computing provider. However, the latter agreement was terminated at the same time as Stargate’s announcement.
In Davos, Nadella also expressed doubts about Stargate’s issues and referred to the planned investments by Microsoft in the amount of $ 80 billion.
“I only know that I’m good for my $ 80 billion,” he said later answer On Musk on the social media platform
Stargate is only the latest example of an infrastructure setback for data centers in the United States, which is preparing for the next stage of the AI economic boom. Musks Xai built one Supercomputer called “Colossus” With 100,000 connected Nvidia chips last year in just three months and promised to increase the number to ten times.
Blackrock and Microsoft prepare the introduction of a 30 billion dollar Ki investment fund for the construction of data centers and energy projects in order to meet the growing requirements of the technology sector. On Friday, meta boss Mark Zuckerberg said that the company would invest between $ 60 and $ 65 billion in the capital infrastructure this year and at the same time expand its AI teams.
“I had customer discussions in all industries continuously. I don’t think there is a single CEO that I have spoken to who doesn’t know that he has to use AI, ”said Friar von Openaai. “AI is not only on the agenda; It is the agenda. It is no longer just an abstract concept or a futuristic vision. It’s here. “
Additional reporting from Harriet Agnew in Davos
