Google’s Former CEO Kinda Doesn’t Want An All-Out AI War

The implementation of artificial intelligence has brought big advances to the tech industry. However, in recent years, some prominent names in the segment have shared their views regarding the “aggressive” development of the technology. In line with that, Eric Schmidt, former Google CEO, is warning, along with other experts, about the risks of pursuing a “Manhattan Project” for AGIs.

Former Google CEO and other experts share their vision on a US Manhattan Project for AGI

Eric Schmidt, Alexandr Wang—Scale AI CEO—and Dan Hendrycks—Center for AI Safety Director—signed a policy paper where they warn about potential retaliation against the United States in a scenario where they try to lead the technology of AGIs (AI agents). According to the paper “Superintelligence Strategy,” technologically powerful countries like China could respond to the United States in the form of potential cyber-attacks. These situations would lead to the destabilization of international relations and escalations in conflicts.

“[A] Manhattan Project [for AGI] assumes that rivals will acquiesce to an enduring imbalance or omnicide rather than move to prevent it,” reads the paper. “What begins as a push for a superweapon and global control risks prompting hostile countermeasures and escalating tensions, thereby undermining the very stability of the strategy purports to secure.”

The paper appears to refer to an American initiative that suggested a kind of “Manhattan Project” for the development of AGIs. The program would be similar to the one that led to the creation of the atomic bomb decades ago. Energy Chris Wright, US Secretary of Energy, recently expressed the idea together with Greg Brockman, OpenAI co-founder, on a supercomputer site.

This is not the first time that Eric Schmidt has warned about the risks of AI in today’s world. The former Google CEO voiced his concerns in mid-February regarding the potential use of AI platforms for biological weapons-based terrorist attacks.

Paper suggests a defensive stance by disabling “threatening AI projects” through cyberattacks

The paper suggests that the US should take a defensive stance regarding AGIs. That is, keep an eye on the development of the technology by rival countries. The co-signers suggest that the US should disable “threatening AI projects” if detected. They say that the country should “expand [its] arsenal of cyberattacks” to prepare for such situations. They are also in favor of the US limiting access to advanced AI chips and its open-source AI models.

The view is a bit fatalistic but in line with what Schmidt has expressed recently. However, the paper contradicts some of the former Google CEO’s advice. Previously, he advocated for the US to actively compete for dominance in the AI ​​field. But when talking about AGIs, Eric Schmidt seems to advocate a more defensive than aggressive stance.