OpenAI’s AI challenges a scientific hypothesis that had stood for nearly 30 years
OpenAI said its artificial intelligence had found a way to disprove one of computer science’s key hypotheses—the exponential time hypothesis (ETH).
Soon afterward, Anthropic presented a similar study, Business Insider reports.
The hypothesis, formulated in 1999, posits that solving certain complex computational problems requires an amount of time that grows rapidly as the volume of data increases. The AI proposed an algorithm theoretically capable of solving such problems faster than previously thought possible.
If the results are confirmed, the discovery could change scientists’ understanding of the limits of computers’ computational capabilities and pave the way for the development of more efficient algorithms.
However, independent verification of the evidence is still ongoing. Experts emphasize that it is too early to speak of a practical breakthrough.
