OpenAI released solutions to three complex mathematical problems, including a refutation of the Exponential Time Hypothesis (ETH).
The release outlines algorithms solving difficult logic puzzles in sub-exponential time, sparking analysis and response across the computer science community.
This theory, published in 1999 by Russell Imbaliajo and Ramamohan Paturi, has been a long-standing assumption about which problems are considered hard for computers. According to the theory, as the variables in certain complex logic puzzles grow larger, faster algorithms take much longer to find solutions.
Both OpenAI and Anthropic claim that their AI models created algorithms that circumvent this rule and solve puzzles much faster, which could potentially debunk the theory. Levent Alpoge, a mathematician at Anthropic, suggested that OpenAI was building on previous work, while Anthropic released its own separate paper on the same topic shortly after.
How did the academic community react?
OpenAI divided the work differently by discipline: it gave the material to mathematicians through a software development platform, while creating working groups for computer scientists and physicists.
↳
READ ALSO
HACKS.GR / RELATED
AI Unlocked
Former Meta CEO leaves and builds rival for Office and Google Workspace
↗
Ryan Williams, a professor at MIT, has taken on the task of organizing the computer science industry's response. Although he had previously criticized similar publications by OpenAI as putting too much pressure on mathematicians, he has now agreed to work with artificial intelligence companies to interpret the solutions together with the creators of the original theory.
An open letter has been published asking for comments from scientists and a discussion about the contribution of artificial intelligence to their work. Ryan Williams noted that, after these findings, computer scientists may need to consider more seriously the possibility that one of the greatest unsolved problems in computing, the equation of two basic complexity classes, may finally be valid.
What does this mean for security and cryptographic systems?
Improved algorithms could theoretically affect digital encryption systems that rely on specific key sizes, making them considered broken.
However, Dan Roberts, chief scientist at OpenAI, has looked into whether any security systems are affected and said he is fairly confident that no significant problem is occurring.
G
HACKS.GR / GOOGLE SEARCH
Hacks.gr on Google
Add us to your preferred sources for more updates from Hacks.gr.
Preferred source
↗
end of article
Top comments (0)