- The former CEO of Google, Eric Schmidt, co-wrote a newspaper warning the United States of the dangers of an AI Manhattan project.
- In the newspaper, Schmidt, Dan Hendrycks and Alexandr Wang are growing for a more defensive approach.
- The authors suggest rival American sabotage projects, rather than advancing the border of AI alone.
Some of the biggest names in AI Tech say that a “Manhattan project” of the AI could have a destabilized effect on the United States, rather than helping to safeguard it.
The terrible warning came from the former CEO of Google, Eric Schmidt, the Director of Security of the Center for AI, Dan Hendrycks, and the CEO of the Alexandr Wang scale. They co-author a policy document entitled “Superintelligence Strategy” published on Wednesday.
In the article, tech ticartts urged the United States to stay away from an aggressive thrust to develop an AI supentinant, or acted, which, according to the authors, could cause international reprisals. China, in particular, “would not be inactive” while the United States worked to update AG and “risk loss of control”, they write.
The authors write that circumstances similar to the nuclear arms race which arose from the Manhattan project – a secret initiative which ended with the creation of the first atom bomb – developed around the AI border.
In November 2024, for example, a Bipartite Congress Committee called for a “Projectan-Like” program in Manhattan, dedicated to pump funds in initiatives that could help the United States beat China in the race in AG. And only a few days before the authors published their article, the American Energy Secretary, Chris Wright, said that the country was already “at the start of a new Manhattan project”.
“The Manhattan project assumes that the rivals will acquire an imbalance or a sustainable omnicide rather than moving to prevent it,” write the authors. “What starts as a push for a super-army and a world control is likely to cause hostile countermeasures and climbing tensions, thus undergoing stability even that the strategy claims to secure.”
It is not only the government that subsidizes the progress of AI, according to Schmidt, Hendrycks and Wang – private companies develop their own “Manhattan projects”. Demis Hassabis, CEO of Google Deepmind, said that he was losing sleep on the possibility of finding himself like Robert Oppenheimer.
“Currently, a similar emergency is obvious in the world effort to conduct AI, with investments in the training of AI on doubling each year for almost the last decade,” say the authors. “Several” AI Manhattan projects “aimed at possibly building a superintelligence are already underway, funded by many most powerful companies in the world.”
The authors argue that the United States is already operating in similar conditions for mutually assured destruction, which refers to the idea that no nation with nuclear weapons will use its arsenal against another, for fear of remuneration. They write that another effort to control the space of AI could cause reprisals of the rival global powers.
Instead, the document suggests that the United States could benefit from a more defensive approach – sabotaging “destabilizing” AI projects via methods such as cyber attacks, rather than rushing to perfect their own.
In order to address “rival states, thug actors and the risk of losing control” at the same time, the authors proposed a triple strategy. Dissusing via sabotage, restrict access to fleas and “weapon infiliation systems” to “rogue actors” and guarantee access to ia fleas via domestic manufacturing.
Overall, Schmidt, Hendrycks and Wang are growing in balance, rather than what they call the strategy “move quickly and break things”. They argue that the United States has the opportunity to take a step back from the emergency rush of the arms race and to move on to a more defensive strategy.
“By methodically forcing the most destabilizing movements, states can guide AI to unprecedented advantages rather than risking that it becomes a ruin catalyst,” write the authors.
businessinsider
