Schmidt Opposes a 'Manhattan Project' Approach to AGI Development

Image: Gemini Imagen 3

• A policy paper authored by Eric Schmidt (former Google CEO), Alexandr Wang (Scale AI CEO), and Dan Hendrycks (Center for AI Safety Director) argues against a rapid, government-led push to develop Artificial General Intelligence (AGI), rejecting the 'Manhattan Project' model.

• The authors contend that such an aggressive pursuit of AGI dominance by the U.S. risks provoking a strong response from China, potentially leading to cyberattacks and escalating international tensions.

• They introduce the concept of 'Mutual Assured AI Malfunction' (MAIM), suggesting proactive disabling of threatening AI projects as a deterrent, rather than a race for supremacy.

• Instead of focusing on winning an AGI race, the paper proposes a defensive strategy that includes expanding cyberattack capabilities to disable adversarial AI projects and limiting access to crucial resources like advanced AI chips and open-source models.

• This measured approach contrasts with two prevailing viewpoints: the 'doomers' who advocate for slowing AI progress and the 'ostriches' who believe in accelerating development regardless of risks.

• This stance represents a shift for Schmidt, who previously advocated for aggressive competition with China in AI development, highlighting the evolving understanding of the strategic implications of AGI and the need for a cautious, defensive approach.

https://techcrunch.com/2025/03/05/eric-schmidt-argues-against-a-manhattan-project-for-agi/