image copyrightReuters

Levent Alfoege, a mathematician and employee of the artificial intelligence (AI) company Antropic, claimed to have proven the 'Riemann hypothesis', a representative difficult problem in the mathematics world, using the company's AI chatbot.

On the 20th (local time), a day after Antropic was listed on the U.S. stock market, Alfoege announced that he had completed a detailed proof of the Riemann hypothesis using the AI ​​chatbot 'Claude' over a period of about six weeks.

If this proof is proven correct, it could go down as one of the most important achievements in the history of mathematics. Alfoege described it as “the beginning of something much greater for science as a whole.”

This claim comes as AI companies, including OpenAI, have recently announced that they have solved difficult problems in advanced mathematics that mathematicians have been unable to solve for decades.

Antropic's AI chatbot 'Claude' model, which has not been released to the public, was used to prove the Riemann hypothesis announced on the 20th.

The Riemann Hypothesis is a approximately 170-year-old mathematical problem involving prime numbers, and is considered one of the most famous difficult problems in mathematics due to its complexity.

This hypothesis, named after German mathematician Bernhard Riemann, states that there is a certain rule for the distribution of prime numbers on a number line.

Since prime numbers form the basis of higher mathematics, new discoveries related to them can have a broad impact on various fields of mathematics.

End of TOP News

Alföge said he contacted Antropic CEO Dario Amodei on the 15th (local time) and asked directly how to announce this discovery.

Regarding the timing of Antropic's stock market listing, he said, "The timing of the announcement was really bad," and added, "If I could have done it, I would have finished all the work much earlier."

He added, "I am entirely responsible for the decision to make the announcement because I felt I could not wait any longer. It is clear that there may be a cynical view regarding the timing of the announcement, but this is not true."

The proof released by Alföge was written in 'Lean', a computer language that verifies the accuracy of mathematical proofs, and the total length is over 250,000 pages.

Renowned mathematician Terrence Tao said on the 20th that it will take a considerable amount of time to complete the verification of this proof.

“The process of understanding and reviewing the proof will likely take time,” Tao said, “but at least this time we can communicate directly with the human author who can help with verification.”

Previously, in major mathematical proofs conducted using AI this year, cases in which human researchers were directly involved were not clearly revealed.

OpenAI claimed in September that its employees had solved part of a notorious math problem in just 88 hours using an AI model that had not been released to the public.

The problem is related to the ‘Navier-Stokes equations’ that describe the movement of fluids such as liquids and gases.

At the time, as part of an AI training experiment, more than 10,000 'AI agents', computer programs that operate autonomously to a certain degree, were used to solve problems.

However, the accuracy of the Navier-Stokes equation-related proof released by OpenAI has not yet been verified. As a result, criticism was raised in academia and mathematics circles regarding the use of AI to solve difficult mathematical problems.

Antropic has also announced other research achievements and problem-solving cases in the fields of mathematics and physics.

Some mathematicians have criticized the practice of AI companies advertising that they have solved major mathematical problems to show off the technological prowess of their commercial products, saying it is “harmful to the discipline of mathematics.”

In a joint letter, they said, “We are witnessing a situation where the results of AI use are inconsistent with the original purpose and threaten intellectual activity in general.”

A month later, OpenAI announced that it had found solutions or proofs or had made significant progress for 372 unsolved high school mathematics problems.

However, it is expected that it will take several months or longer for experts to verify whether these claims are valid.

Learn more

7 founders who will become worth tens of trillions of won through Antropic's listing

‘Semu cat, Lit cat’… Controversy over AI cheating, can we no longer trust national exams?

The 'NVIDIA spy scandal' that broke out before the US-China meeting... Will it become an obstacle to China's AI ambitions?

video

Video, 'Nobody was punished' for alleged gang rape at Cornell University, broadcast length 1,34

Video, 'Irreversible water temperature' How hot have the seas of Ulleungdo and Dokdo become, broadcast length 5,34

Video, North Korean defector women talk about dating and sex in North Korea, broadcast length 3,59

Video, from before to after the 'K-pop craze'...the path taken by female musicians, broadcast length 12,14

top news

A year in which AI solved math problems... Antropic, has it solved the biggest challenge?

GTA6, highest rating... Received a ‘masterpiece’ rating

‘The highest peak of human health’... The secret of a super potty worker with healthy stools

this time issue

Most viewed news