Artificial intelligence does very quickly what takes us a long time. Its impact is that of the robotization of intellectual processes. While computers have revolutionized the world of work for blue-collar workers, AI will revolutionize it for white-collar workers. We are already witnessing a series of disasters, not because of AI itself, but because of those who manipulate it. We must not test the quality of our AI, but the intelligence of those who trust it.
Our media are gripped by a debate about the value and dangers of Artificial Intelligence (AI). However, under the influence of science fiction writer Eliezer Yudkowsky, now a Silicon Valley guru, they focus on a mythical approach: that of a machine gaining the upper hand over humanity and destroying it. This is merely a rehash of the themes of the golem or Frankenstein. But since artificial intelligence cannot create, but only reproduce at very high speed, this danger is irrelevant.
Epsoak Epsom Salt - 5 ...
Buy New $12.99 ($0.16 / Ounce)
(as of 01:05 UTC - Details)
The danger we face right now is becoming fascinated by machines. Many people develop emotional attachments to chatbots. They tell them what they want to hear and encourage their flaws. This is a problem already well known to psychologists.
The use of AI by military commands began in the 1990s, during the Gulf War, to solve logistical problems by processing a large number of factors simultaneously. Gradually, numerous weapons were equipped with AI. Problems only arose when AI was relied upon to make real-time decisions, that is, since 2024.
France decided to establish a Ministerial Agency for Artificial Intelligence in Defense (AMIAD) in March 2024, although it was only created two months later. This agency set itself a limit: never to entrust the AI of a weapon with the ability to choose to kill.
Conversely, the Pentagon, which does not share the same ethical principles, has just created “Project Meridian” [ 1 ] . Following its conflict with Anthropic over the limits the company intended to place on the use of its AI, Claude, the Pentagon does not want to be accountable to anyone. It has therefore tasked Elon Musk and Palmer Luckey (the creator of virtual reality headsets) with envisioning future developments.
Several cases of poisoning have been recorded. They all led to disasters, even though they were easily avoidable. In each case, those in charge trusted the AI to the point of asking it what it couldn’t possibly know.
What’s striking is that very high-ranking figures, unsure of their own judgment, were easily manipulated by AI. Conversely, people who are “down-to-earth” immediately become skeptical when an AI spouts nonsense.
• During the response to Operation Al-Aqsa Deluge (October 7, 2023), the Israel Defense Forces (IDF) used AI to determine targets for their bombings. The AIs were set to assume that collateral damage was negligible. Initially, the AIs were configured to risk no more than two civilian casualties per target destroyed. This threshold was quickly raised until the AIs disregarded it altogether.
This shift in focus was reflected in a change of AI. The system was replaced by Habsora and Lavender. The objective was no longer to assess potential collateral damage, but to concentrate on the 37,000 suspected Hamas and Islamic Jihad militants.
The officers who authorized the firing were only concerned with verifying that the targets were indeed members of Palestinian organizations. They did not intentionally commit genocide. They simply allowed AI to do it, without even realizing it. By the time they opened their eyes, it was already too late [ 2 ] .
What distinguishes a civilized army from a band of barbarians is its commitment to protecting civilians. Certainly, members of the government repeatedly told them that all Palestinians were terrorists, but the prevailing savagery did not absolve them of their responsibility. Shocking as it may seem, they inadvertently committed genocide.
Dr. Bronner’s Pe...
Buy New $32.99
(as of 03:31 UTC - Details)
In these cases, there was no error on the part of either the AI or the operators. The crime was committed through negligence. According to journalistic investigations, it appears that the younger and more inexperienced the officers were, the more easily they forgot about reality and caused the massacre of civilians.
• During the “12-Day War” (June 13–24, 2025), Western powers consulted the International Atomic Energy Agency (IAEA) to ascertain the status of Iran’s nuclear research. Its director, Rafael Grossi, then consulted his version of Mosaic, an AI developed by Databricks. It confirmed that Tehran was suddenly on the verge of developing an atomic bomb; an opinion that was widely disputed by his own inspectors [ 3 ] . Mr. Grossi’s impulsive statements nearly provoked a nuclear response from the United States.
Databriks is headed by Iranian-Swedish-American Ali Ghodsi. Although he is not involved in politics, he is a member of the Israel Collaboration Network (ICON), a network of computer scientists. The version of Mosaic installed in Vienna for the IAEA was installed by a group of Israeli engineers.
There are two ways to interpret this event. Either one believes that Rafael Grossi was manipulated by Mossad, or one observes that he was impressionable enough to follow a machine rather than his own specialists.
• In December 2025, Donald Trump questioned Grok, Elon Musk’s AI, about how Venezuelans would react to a potential kidnapping of President Nicolás Maduro. This was a planned operation discussed by the CIA and SouthCom. The AI’s response, according to Time Magazine , was that the Venezuelan people would be delighted. Based solely on this opinion, Donald Trump allegedly launched the operation [ 4 ] .
Why seek other opinions when an AI can synthesize everyone’s thoughts? Simply because an AI only formulates the average of the thoughts it scans. And no average is the truth. AIs are just algorithms. They don’t think; they simply reformulate the thoughts we want to explore. Someone familiar with the Venezuelan population would have responded with the complexity of the country’s culture in mind. They certainly wouldn’t have agreed or disagreed with President Trump’s question, but would have emphasized that Venezuelans’ reactions would depend on their social and regional background, as well as the circumstances of the event.
Recently, Russian President Vladimir Putin explained that the problems encountered with AI are not new. They already existed with far less sophisticated machines.
Pure Encapsulations Al...
Buy New $67.50
(as of 03:31 UTC - Details)
He told the Valdai Club that, during the Cold War, in 1983, Soviet satellites had observed the United States launching five nuclear missiles. Colonel Stanislav Petrov, who was in charge of the Soviet response, immediately questioned what the machines were telling him. He first verified the information before ordering a counterattack. He was right to do so. The machines were wrong; the US had not launched a nuclear war, and neither had he . [ 5 ]
It is indeed the same problem. We are still the masters of the machines, but our inattention, our incompetence or our pride can cause us to make small mistakes which, in the very short term, will have terrifying consequences.
In truth, no one can be manipulated by an AI, simply because an AI cannot have intention, will of its own, but we all risk, through laziness, giving credence to what an insufficiently calibrated machine tells us.
—
You may freely reproduce articles from the Voltaire Network provided you cite the source and do not modify them or use them for commercial purposes ( CC BY-NC-ND license ).
The Use and Misuse of Science