OpenAI chief executive Sam Altman said society should accept that artificial intelligence will cause some “bad things” in exchange for the technology’s benefits and the freedom to use it. His comments came as investigations described AI-linked systems being used in military decisions with potentially lethal consequences.
An AI-linked system called Maven, made by Palantir, was used to help target a school in Gaza, where an investigation concluded that the United States was probably responsible for the strike. Unnamed defense officials said some personnel at the US Central Command relied too heavily on artificial intelligence embedded in Maven Smart System, which the Defense Department had rapidly made a central part of its military operations.
In a separate incident, a US military chatbot incorrectly assessed that a Chinese ship in Middle Eastern waters was carrying components for a nuclear weapons program. US aircraft launched an interception mission, and personnel prepared to board the vessel before a closer review exposed errors in the chatbot’s analysis. The analyst who used the chatbot for the initial assessment also used AI to write the report that drove the response, compounding the original mistake.
The contrasting examples show an implication of Altman’s formulation: the consequences described as “bad things” can include decisions involving human lives and international conflict, rather than only minor consumer errors.