New Delhi: While the general consensus on AI “hallucinations” – the generation of plausible but false information – are concerning, scientists are increasingly recognising their potential for driving innovation and discovery, report has said, adding that these unexpected outputs from AI models are proving surprisingly useful in fields ranging from medicine to climate science.
According to a report in The New York Times, AI hallucinations, or inaccurate or misleading outputs can spark new ideas and accelerate the scientific process. AI hallucinations arise when generative models are trained on a specific subject and then allowed to rework that information, sometimes producing unexpected and even surreal results.
“The public thinks it’s all bad. But it’s actually giving scientists new ideas. It’s giving them the chance to explore ideas they might not have thought about otherwise,” Amy McGovern, a computer scientist and director of a federal AI institute, was quoted as saying.
AI hallucinations speed up hypotheses
According to the scientists, this ‘creative aspect’ of AI is proving valuable particularly in the early stages of scientific discovery, where hunches and guesswork often play a crucial role. AI hallucinations can speed up the process of generating and testing hypotheses, leading to faster breakthroughs.
James J. Collins, an MIT professor, recently praised AI hallucinations for accelerating his research into novel antibiotics.
“We’re exploring. We’re asking the models to come up with completely new molecules,” he said.
The impact of AI hallucinations is evident in the work of Nobel laureate David Baker, who shared the 2024 Nobel Prize in Chemistry for his research on proteins. Baker credited AI hallucinations with helping his lab design millions of brand-new proteins not found in nature, leading to numerous patents and the founding of over 20 biotech companies.