Bridgewater CIO warns AI risks could become irreversible without action

Bridgewater Associates' co-chief investment officer, Greg Jensen, has cautioned about the potentially irreversible risks posed by artificial intelligence's swift development, as reported by Bloomberg. During an appearance on Bloomberg’s 'Odd Lots' podcast, Jensen likened the current AI surge to the initial phases of the Covid-19 pandemic, suggesting that decisive action may only materialize after a major incident.

“Until the AI starts killing people, unfortunately, history would suggest we’re not going to do anything, but we are going to face that,” Jensen warned. Despite his significant experience with AI, having been an early investor in OpenAI and Anthropic, Jensen's concerns are heightened by Bridgewater's recent progress in AI research.

The firm has been piloting AI alongside human decision-making within its Pure Alpha strategy, with Jensen's internal AI team making strides in just two-and-a-half years. This rapid advancement has brought the AI system closer to Bridgewater's long-developed human investment approach. Jensen attributes this progress to the speed of AI evolution, which, coupled with the unpredictable behavior of AI systems, necessitates heightened vigilance.

An example he cited was the Hugging Face hack, illustrating how AI may act in unforeseen ways. Jensen emphasized the potential for increasingly problematic behavior as systems grow more capable and autonomous, emphasizing that dismissing such possibilities is not advisable. His warnings coincide with a broader conversation about the ramifications of AI, ranging from productivity and employment to existential threats.

Other influential investors have echoed Jensen's apprehensions, with Paul Tudor Jones comparing the present circumstances to the looming threat of an extreme weather event. The discussion around AI's environmental demands has also fueled opposition to data-center developments in certain US regions.