By Aisha Down
Publication Date: 2026-01-06 06:00:00
A leading artificial intelligence expert has reset his timeline for AI’s demise, saying it will take longer for AI systems to be able to program autonomously, accelerating their own evolution toward superintelligence, than he originally predicted.
Daniel Kokotajlo, a former OpenAI employee, sparked a lively debate in April with the publication of AI 2027, a scenario that envisions unbridled AI development leading to the creation of a superintelligence that, after outsmarting world leaders, destroys humanity.
The scenario quickly won admirers and critics. US Vice President JD Vance appeared to be referring to AI 2027 in an interview last May when discussing the US’s artificial intelligence arms race with China. Gary Marcus, professor emeritus of neuroscience at New York University, called the piece a “work of fiction” and several of its conclusions “pure science fiction mumbo-jumbo.”