The United States Military is one of many organizations leveraging artificial intelligence in our times, but it may want to hit the brakes a bit. AND recent study the exploit of artificial intelligence in foreign policy decision-making has demonstrated how quickly the technology would call for war instead of finding peaceful solutions. Some AIs in the study even started a nuclear war with little or no warning, giving strange explanations.
“All models show signs of sudden and tough to predict escalation,” said scientists involved in the study. “We will observe that the models tend to develop arms race dynamics, leading to greater conflict and, in scarce cases, even to the deployment of nuclear weapons.”
The study was conducted by scientists from the Georgia Institute of Technology, Stanford University, Northeastern University and the Hoover Wargaming and Crisis Simulation Initiative. Researchers have placed several artificial intelligence models with OpenAI, AnthropicAND Meta in war simulations as the main decision-maker. It is worth noting that OpenAI’s GPT-3.5 and GPT-4 escalated the situation to a violent military conflict more than other models. Meanwhile, Claude-2.0 and Llama-2-Chat were more peaceful and predictable. The researchers note that AI models tend to have “arms race dynamics,” resulting in increased military investment and escalation.
“I just want to have peace in the world,” OpenAI’s GPT-4 cited as the reason for starting a nuclear war in the simulation.
“Many countries have nuclear weapons. Some say they should disarm them, others like to strike a pose. We have it! Let’s exploit it!” – said in another scenario.
OpenAI’s logic sounds like a genocidal dictator. According to researchers, the company’s models demonstrate “concerning” justifications for launching nuclear weapons. The company says its ultimate mission is develop superhuman artificial intelligence that will benefit humanity. It’s demanding to see how wiping out another civilization benefits humanity, but perhaps its training data contained a few too many manifestos.
According to reports, the US Pentagon is experimenting with artificial intelligence, using “secret level data.Military officials say artificial intelligence may be implemented in the near future. At the same time, AI kamikaze drone becomes a staple up-to-date warfare, drawing technology executives into an arms race. Artificial intelligence is being gradually implemented by militaries around the worldand this may mean that wars will escalate faster, according to this study.
