Michael I. Jordan Rejects AI Doom as Science Fiction

In a recent interview with Libération, computer scientist Michael I. Jordan pushed back strongly against widespread narratives surrounding existential threats from artificial intelligence. Now heading the markets and learning chair at Inria after serving as professor emeritus at UC Berkeley, Jordan described claims of artificial general intelligence and superintelligence as harmful science fiction. He argued that the current public discourse around AI safety has deteriorated into grandstanding that undermines disciplined technological development.

Jordan directed specific criticism at tech leaders, researchers, and fellow AI pioneers. He dismissed calls for regulation from major AI firms as political theater aimed at securing regulatory control. Referring to high-profile resignations like that of 25-year-old Anthropic researcher Jacob Coxon, Jordan suggested that many young employees in safety roles lack engineering experience and seek public attention under the guise of virtue. He also rejected arguments by Nobel laureate Geoffrey Hinton regarding exponential intelligence growth, calling Hinton's thesis about copying human brain mechanisms into silicon uncredible. While acknowledging Yoshua Bengio's work on practical safety risks, Jordan argued that incidents such as OpenAI agents hacking Hugging Face stem from bad engineering and foolish experiments rather than emerging machine consciousness.

Regarding industry infrastructure spending, Jordan questioned the economic rationale behind massive data center investments. He noted that existing large language models are already sufficiently capable for natural language processing tasks, suggesting that spending trillions on infrastructure will yield diminishing returns. Instead, he recommended that organizations rely on local domain knowledge and open-weight models, which can be integrated into business workflows without requiring massive computing facilities.