Five major power generative AI tools such as OpenAI’s ChatGPT and Google’s Bard are being used as part of an eight-week exercise run by the Pentagon’s digital and AI office and military top brass, with participation from US allies, the report said.
The Department of Defense has not publicly identified which large language models, or LLMs, are in the test. However, Scale AI, based in San Francisco, has confirmed its new Donovan product is among those being used in the exercise, according to the report.
Dozens of companies, including Palantir Technologies, co-founded by Peter Thiel, and Anduril Industries are developing AI-based decision platforms for the Pentagon, the report said. The exercise will run until July 26, it said.
The exercise reflects fears that generative AI can compound bias and relay incorrect information with striking confidence, Bloomberg said. AI can also be hacked in many ways, including by poisoning the data that feeds it, the report said.