A recent report by leading AI researchers, including Daniel Kokotajlo (formerly of OpenAI), Scott Alexander, and Eli Lifland, projects that artificial intelligence could outperform humans in nearly all tasks by 2027, with the advent of superintelligence following shortly thereafter.
The report, titled “AI 2027,” outlines a scenario where AI systems rapidly evolve from superhuman coding capabilities to surpassing human performance across most jobs within months. By 2028, these systems could achieve superintelligence, potentially reshaping science, industry, and society on a scale comparable to the Industrial Revolution.
The authors present two potential futures: one where development slows to prioritize safety and understanding, and another where global competition accelerates progress with little time for precaution. They emphasize that while this future is not guaranteed, it is plausible enough to warrant immediate planning and consideration.
This projection aligns with other expert analyses. For instance, a comprehensive essay by Leo Aschenbrenner likens the urgency of preparing for superintelligent machines to the Manhattan Project, advocating for significant investments in AI safety and alignment.
As AI continues to advance at an unprecedented pace, the global community faces critical decisions on how to navigate the potential benefits and risks associated with these transformative technologies.
For a more in-depth understanding, you can watch the discussion between Daniel Kokotajlo and Eli Lifland on their AI 2027 report:
