Daniel Kokotajlo, a former OpenAI employee and author of the AI 2027 scenario, has revised his timeline for artificial general intelligence (AGI) and its potential to destroy humanity. In an update, Kokotajlo now predicts that AI will achieve fully autonomous coding in the early 2030s, rather than 2027, pushing the arrival of superintelligence to 2034.
The AI 2027 scenario, released in April, sparked debate by suggesting that unchecked AI development could lead to a superintelligence that outfoxes world leaders and destroys humanity by mid-2030. The scenario gained attention from US Vice-President JD Vance, but was criticised by Gary Marcus, an emeritus professor at New York University, who called it a “work of fiction”.
Kokotajlo’s revised forecast reflects growing doubts about the imminence of AGI. Malcolm Murray, an AI risk management expert, noted that “a lot of other people have been pushing their timelines further out in the past year, as they realise how jagged AI performance is”. Henry Papadatos, executive director of SaferAI, added that the term AGI “does not mean as much” now that AI systems are already quite general.
The new timeline does not include a specific prediction for when AI might destroy humanity. Kokotajlo wrote on X: “Things seem to be going somewhat slower than the AI 2027 scenario. Our timelines were longer than 2027 when we published and now they are a bit longer still.”
Despite the delay, leading AI companies continue to pursue automated AI research. OpenAI CEO Sam Altman said in October that having an automated AI researcher by March 2028 was an “internal goal”, though he acknowledged the company “may totally fail at this goal”.
Andrea Castagna, an AI policy researcher, cautioned that dramatic AGI timelines overlook real-world complexities. “The more we develop AI, the more we see that the world is not science fiction,” he said.
