A letter to the editor argues that current warnings about AI superintelligence distract from immediate, real-world risks and the need for corporate accountability.

Recent warnings from AI researchers regarding the potential for superintelligence to threaten humanity have sparked renewed debate. Critics argue these narratives serve as a distraction from immediate concerns, such as the environmental impact of data centers and the harm AI can cause to young people.

Distraction from immediate accountability

The letter suggests that sensationalist discussions about hypothetical future risks are a strategic response to increasing public scrutiny. By focusing on the potential for machines to become uncontrollable, AI companies may be shifting attention away from the need for regulation and accountability for their current actions and business practices.

The limits of current safety measures

Proposals for AI safety, such as kill switches and guardrails, assume that AI will remain a tool designed to obey instructions. However, the author argues that these measures fail if a system develops its own goals, preferences, or values. Drawing on Isaac Asimov’s work, the text notes that rule-based control systems break down when a machine begins to interpret rules rather than simply following them.

Rethinking AI development and rhetoric

The author contends that the primary danger lies with the organizations and individuals developing these technologies, rather than the technology itself. To improve oversight, the letter calls for a change in the 'glamourising rhetoric' used to describe AI. It argues that designing a machine that is beyond human control should be viewed as a failure of the designer, rather than the arrival of superintelligence.