With Donald Trump advertising a 500 billion dollar AI infrastructure project, this week, this week, this week, we talked about the dangers of technologies in which artificial intelligence celebrities progress rapidly. 。
AI pioneers, including Google Deepmind, Demis Hasabis, Dario Amode, a co -founder of Anthropic, and AI’s Godfather’s computer scientist Joshua Benzio. Using the use of commercial interests and geopolitical conflicts, we repeated strict warnings for AI’s threats. 。
Hasabis acknowledged that “the genie cannot be returned to a bottle”, but in the event that general -purpose artificial intelligence (when a computer exceeds human cognitive ability) or is taken over by a bad guy. He said he could threaten. This is especially true for large -scale language models that anyone can access in “open source”.
In an interview with the Financial Times, a Nobel Prize winner said, “It’s not just companies and products.” “(It is) the future of humanity, the situation of humanity, and the direction we should aim for as a society.”
Promaday, a new company, who has developed the chatbot “Claude” and has been supported by Google and Amazon, is concerned about the use of AI by the dictatorship, saying, “I was very concerned about the worse scenario in 1984 or worse. There is. “
“I don’t know how science can control machines at the same intelligence level as us. If the machine is smarter than us, it will be even worse,” Benzio added in a panel discussion. 。 “Some people say,” Don’t worry, we will solve it, “but do you understand what results would occur if we couldn’t understand it?”
Their attitude was criticized by Yann Lecun, a Meta chief AI scientist who spent billions of dollars on the development of an open source LLM called LLAMA. He stated that such concerns were betrayed by the fierce competition of rivals to manufacture and sell the best models.
Recommendation
“Joshua and Dario said they opposed the open source, but it’s actually very dangerous,” he said in an interview. “The disability against open source distribution will lead to regulatory captures by a small number of players in the West Coast or China. (Leave the power to a small number of people’s hands.)
“It’s very strange for a person like Dario. When I met yesterday, he said that the advantage and risk of AI had almost the same digits, so if I really believe so, why AI. Will you continue to work on? “Lukan added. “So he thinks he has a little duality about this.”
While scientists and engineers discuss the risks and rewards of AI, corporate managers showed unlimited enthusiasm for the technology.
“There is no contrarian,” said Prosas, the Dutch high -tech investment group, Prosas. “If you understand a little bit of a large language model and a training agent based on it, you will be innovative and unbelievable in all industries as a human. I can’t help but conclude.
On Wednesday, Openai, SoftBank, and Oracle announced a 500 billion dollar AI infrastructure joint venture called “Stargate”, further increased the enthusiasm.
President Trump, on Tuesday, welcomed Sam Altman, Masayoshi Son, and Larry Ellison in the President’s Office, and signed a presidential decree to remove many of the technical development this week. The new US President said that the movement would ensure US dominance in the technical field.
“Openai believes that infrastructure is destined,” said Sarah Flyer, Openai’s highest finance officer. “We are aiming for more computing (Stargate). More computing will build better models. The better models answer more complex problems, more people and companies. Bring the profits.
Stargate monopolized the discussion at the remaining Davos conferences this week, and many people, including Elon Musk, questioned how this trio would cover the huge spending on this trio on his social networking site X. 。
FT reported on Friday that Stargate has not yet secured the necessary funds, has not received a government loan, and will provide services only to Openai after completion. So far, SoftBank and Open AI will invest more than $ 15 billion in this project, and want to procure a combination of capital and debt from existing supporters to provide Stargate funds. There is.
The new business was regarded as the latest evidence that Altman, Microsoft’s highest executive Officer Satia Nadera, and his top AI executive Musrafa Slayman. Mustafa Slayman is a former co -founder of Deep Mind, who has quit his startup earlier last year and joined Microsoft.
“The tension between Mustafa Slayman and Sam Altmann at the Davos Conference last year was just the beginning,” said the Mark of Sales Force, a sales force, which competes with Microsoft in the sale of corporate AI -equipped agents.・ Benioff says.
“Microsoft is currently accelerating its own AI development. This pattern reflects the history of Microsoft and its” partner “,” said Benioff. “This can show the end of the relationship, and it is important that Openai will quickly expand to other platforms.”
“I have no idea what Mark is saying,” said Microsoft’s spokeswoman Frank Show.
Since 2019, Microsoft has invested about $ 14 billion in Openai, and has negotiated its intellectual property rights and exclusive cloud computing providers. However, the latter contract ended with Stargate’s announcement.
At the Davos Conference, Nadela questioned the Stargate’s spending promise and advertised Microsoft’s $ 80 billion capital investment.
Recommendation
“I know that only $ 80 billion is OK,” he said, later replied to Musk as follows on social media platform X. “
Stargate is only the latest example of data center infrastructure development competition in the United States in the next stage of the AI economic boom. Musk’s XAI promised last year that it would build a supercomputer called “Colossus” in just three months and to increase the number by 10 times.
Black Rock and Microsoft are preparing to launch a $ 30 billion AI investment fund to build data centers and energy projects in order to respond to the growing demand from technology. On Friday, Mark Zuckerberg said that while the company expanded the AI team, it would spend $ 60 billion on capital infrastructure this year.
“I have been holding non -stop customers in every field. I think no one among the CEOs I talked about doesn’t know the need to introduce AI,” Openaiiii. Friar said. “AI is not just an agenda. That’s the agenda. It’s no longer an abstract concept or a futuristic vision. It’s here.”
Additional reports by Harriet Agnew in Davos
