An AI system will collaborate with humans to achieve a Nobel Prize-winning discovery within the next 12 months, according to Jack Clark, co-founder of Anthropic. Speaking at Oxford University, Clark described a "vertiginous sense of progress" in AI technology and issued a series of bold predictions regarding the immediate future of the industry.
Clark’s timeline suggests rapid integration across sectors: bipedal robots are expected to assist tradespeople within two years, while companies operated entirely by AI could generate millions in revenue within 18 months. Perhaps most significantly, he predicted that by the end of 2028, AI systems would be capable of designing their own successors, marking a pivotal shift toward autonomous technological evolution.
However, the lecture was balanced by stark warnings about safety. Clark stated there remain plausible scenarios where the technology has a "non-zero chance of killing everyone on the planet," emphasizing that existential risks have not dissipated despite advancements. While Anthropic is best known for its Claude models, Clark referenced a newer version called Mythos, which demonstrated alarming proficiency in exploiting cybersecurity vulnerabilities.
Addressing the student body, Clark admitted it would be preferable for humanity to slow down development to allow the species more time to adapt to these powers. He noted, however, that this is unlikely given the "breakneck development" driven by various actors and nations. He lamented that commercial and geopolitical rivalries often drown out larger existential concerns, leaving humanity in a state of denial. Comparing the situation to pandemic unpreparedness, he warned that allowing synthetic intelligence to multiply unchecked would eventually force humanity into a purely reactive stance.
Critics of frontier AI firms like Anthropic and OpenAI expressed concerns over the concentration of power, fearing a "single point of failure" in global systems due to over-reliance on a few dominant models. Furthermore, Prof. Edward Harcourt, director of the Institute for Ethics in AI, warned that the rise of AI-driven automation risks creating "cognitive atrophy," potentially weakening human judgment and decision-making capabilities. He advocated for AI models that prompt human engagement rather than simply providing automated conclusions.
