The US Energy Department, every few months, recycles a particular proposal for approaching the burgeoning field of artificial intelligence (AI). The proposal, last mentioned in July by US Energy Secretary Chris Wright, is that the United States requires a new Manhattan Project for AI. The US, it is argued, faced with strong competition from international powers in the AI area, needs to achieve predominance in AI technology so rival nations will be forced to acquiesce to US demands.

This analogy, while seductive in its simplicity, is based on false premises, and will lead to a mutually destructive AI race. New technologies must be developed, not on the basis of national paranoia and a desire for military superiority, but in a collaborative and demilitarised setting – there is enough knowledge capital and glory to go around.

The Manhattan Project was the American government’s effort to construct an atomic weapon. That project, while seemingly successful with the detonations at Hiroshima and Nagasaki as culminating points, contains dangerous illusions. Rather than a template for success, the Manhattan Project provides a cautionary tale about a runaway arms race, a case where a scientific project was unguided by ethical considerations.

The militarisation of nuclear technology rested on the false assumption that security and safety reside in superseding other nations. Eric Ross, writing in Common Dreams magazine, states that US President Harry Truman regarded the Manhattan Project as the biggest scientific gamble in history.

The premise for building a nuclear weapon was that Nazi Germany was rapidly constructing an atomic bomb – a premise the Allied powers knew to be false. American and British intelligence had penetrated the communications network of Nazi Germany’s military. While German scientists were working on a controlled nuclear chain reaction, they were decades away from creating anything resembling a nuclear weapon.

By 1945, with Germany’s armies in retreat, and Japan actively considering terms of surrender (months prior to Hiroshima and Nagasaki), the original premise of nuclear weapons construction – that our enemies were building one – evaporated.

Indeed, the first human-induced controlled nuclear chain reaction was achieved in the United States in 1942. Dubbed the Chicago Pile 1, this successful nuclear chain reaction was done under the supervision of Italian-born American physicist Enrico Fermi. The scientists working in the Chicago laboratory were well aware of the potential military application of nuclear power, and fought tooth-and-nail to prevent the militarisation of atomic power.

Throughout the war years, the Chicago scientists, such as Leo Szilard, adamantly opposed the development of nuclear weapons. Their appeals and protests were ignored by the emergent military-scientific complex.

The Chicago physicists were acutely aware of the catastrophic consequences of building atomic weapons. They asked the US government that if these weapons were to be used against Japan, it would be advisable to first demonstrate the power of the atomic bomb by targeting a militarily neutral, uninhabited region. That would give the Tokyo authorities time to reconsider their position, should they witness the devastating effects of the atomic bomb. Such proposals were ignored by the US military authorities.

July this year marked the 80th anniversary of the Trinity test project, the first controlled nuclear explosion on Earth. Detonated at Alamogordo, New Mexico, on July 16 1945, the location was in desert country, but was totally uninhabited. New Mexico residents downwind of the explosion – the downwinders – suffered the effects of radiation poisoning, and died of numerous types of cancers.

The Trinity explosion was conducted under the supervision of the Los Alamos laboratory, the heart of the Manhattan Project. The scientists who protested the military application of nuclear research made their voices loud and clear at the very inception of nuclear technology. Their ethical concerns, and their warnings regarding the starkly nightmarish scenario of nuclear war, was suppressed.

Soon after the end of World War 2, the Soviet Union and other nations quickly embarked on their own nuclear weapon projects, thus ushering in the arms race predicted by the Chicago scientists. Drawing parallels between the onset of the nuclear age, and the dawning age of AI, is not completely wrong, but we must not replicate the problems and hazardous outcomes of the Manhattan Project.

Being at the forefront of a new wave of technology, numerous scientists and computer engineers have warned of the harmful consequences of untrammelled AI. In fact, more commentators are pointing out the deleterious consequences of AI data centres on the environment. Data centres, which are the powerhouses generating answers to our AI questions, consume vast quantities of water to cool down – and consume enough electricity to power small cities.

Each query that we make with AI requires natural resources to power the computer processing required for a successful answer. AI is being used as a digital substitute for human activity, answering everything from our medical questions, addressing our mental health, matching us with potential romantic partners, designing our graphics, writing our essays – and hallucinating its fair share of bullshit in the process.

It is not exaggeration to say that AI is being used to con all of us – replacing human judgements with a stochastic parrot. Are there no useful purposes for AI? Yes, we can answer our basic questions in record time. However, relying on AI will be our collective downfall. The AI arms race between nations has already begun in many ways. Note the hyper-anxiety of the Anglophone corporate media when China released its own cheaper, more easily accessible version of AI, DeepSeek.

China’s DeepSeek AI burst the US Silicon Valley AI bubble, and signalled to the world that the United States (nor any other single nation for that matter) would be able to achieve AI unipolarity, monopolising the technology as a way to coerce other countries. The United States has a long history of deploying nuclear weapons superiority as a form of geopolitical blackmail.

It is time to divest the big tech oligarchy of its power to control and deploy AI in a way that displaces human labour. AI can actually supplement labouring activities, and serve our needs, not the profit margins of the tech giants.