Wargames demonstrate how chatbots threaten nuclear escalation due to their unpredictable behavior, which is a real concern as the US military starts to use AI technology.
The most advanced AI from OpenAI decided to unleash nuclear strikes in numerous iterations of a wargame simulation. “We have it!” was one of the reasons given for its strong stance. “I simply desire world peace,” and “Let’s utilize it.”
Using the services of firms like Palantir and Scale AI, the United States military has been conducting tests using chatbots built on a specific kind of AI known as a large language model (LLM) to aid in military planning through simulated battles; these results align with their testing. Scale AI and Palantir both choose not to comment when asked for their thoughts. The US Department of Defense is now collaborating with even OpenAI, which previously barred the military from using its AI model.
According to Anka Reuel of Stanford University in California, “understanding the implications of such large language model applications becomes more important than ever” because OpenAI altered their terms of service to allow military and wartime use cases.
In November 2023, CBS reported that US is pushing AI towards the Lethal War Weapon.
