Artificial intelligence laboratories OpenAI and Anthropic have triggered an intense debate across the computer science community after releasing conflicting claims regarding one of the field’s foundational assumptions.
On Monday, OpenAI announced that it had punctured the Exponential Time Hypothesis, or ETH, a nearly 30-year-old tenet holding that certain computational logic problems become overwhelmingly time-consuming as their scale increases. Just hours after OpenAI’s disclosure, rival artificial intelligence developer Anthropic released its own research directly refuting ETH.
To evaluate its assertions, OpenAI enlisted a group of outside computer scientists to organize an open review of the claim. However, the participating researchers have publicly emphasized that their involvement does not constitute an endorsement of the company’s findings. If the claims made by the artificial intelligence labs ultimately hold up under scrutiny, they could reopen exploration into increasingly capable algorithms—an area that many computer scientists had previously considered largely exhausted.
The technical announcements have generated extensive discussion across social media platforms and specialized technology circles, drawing reactions from prominent academics, mathematicians, and industry figures.
Ryan Williams, a professor of electrical engineering and computer science at the Massachusetts Institute of Technology, is serving as one of three independent algorithm design experts reviewing OpenAI’s claims. While Williams has not certified the company’s results, he wrote on the social media platform X that verified findings of this nature “should make every computer scientist put at least a little more weight on the possibility of P = NP,” bringing the field closer to another potential $1 million Millennium Prize Problem. From a more practical standpoint, Williams noted that the breakthrough could ultimately yield faster algorithms, adding that he has historically doubted the validity of ETH.
Other prominent academic and industry voices offered varied perspectives on the significance of the announcements. Gary Marcus, an emeritus professor at New York University and a frequent artificial intelligence skeptic, acknowledged on X that the development could represent “a big deal if correct.”
John D. Cook, a Doctor of Philosophy and applied mathematics technology consultant, urged caution in a blog post regarding the broader implications. While describing the claims as impressive, Cook maintained that the developments do not resolve computer science’s most famous open question. The status of P equals NP remains unresolved, Cook wrote, noting that while 3-SAT may not require exponential time, it continues to demand more than polynomial time according to current knowledge.
Joshua Fonseca Rivera, an artificial intelligence safety research fellow at Anthropic, wrote on X that OpenAI’s claim serves as a reminder that human observers may have underestimated machine intelligence. Describing the development through the lens of foundational machine learning lessons, Rivera admitted to a sense of shock despite logical familiarity with rapid technological acceleration.
Timothy Gowers, a professor at the Collège de France and Fields Medal-winning mathematician, offered a direct assessment on X regarding the contradictory claims, calling it “very surprising and consequential that it is false.”
Meanwhile, the announcement also drew lighter commentary from the startup sector. Dan Shipper, chief executive officer at Every, reacted on X by joking about his academic background in light of the rapid theoretical shifts occurring in the industry.
