FDA Bets On AI To Speed Up Drug Approvals

The Food and Drug Administration (FDA) has officially stated its plan to immediately implement Artificial Intelligence across all its centers. This decision follows the successful completion of a new generative AI pilot program designed for its scientific reviewers.

This AI tool is expected to significantly accelerate the FDA's drug review process. The aim is to free up scientists from tedious and repetitive tasks. However, these claims are met with some skepticism, especially considering AI's known issues with generating inaccurate information, or "hallucinating".

Jinzhong Liu, a deputy director in the FDA's Center for Drug Evaluation and Research (CDER), expressed strong support in a statement: "This is a game-changer technology that has enabled me to perform scientific review tasks in minutes that used to take three days."

FDA Commissioner Martin Makary has set an ambitious deadline, directing all FDA centers to achieve full AI integration by June 30.

The agency's announcement elaborated, "By that date, all centers will be operating on a common, secure generative AI system integrated with FDA's internal data platforms."

This news surfaced just one day after Wired reported on discussions between the FDA and OpenAI regarding the agency's AI utilization. Interestingly, the FDA's recent statement does not mention OpenAI or any potential collaboration.

However, sources cited by Wired indicate that a team from OpenAI, the creator of ChatGPT, met with FDA officials and two individuals associated with Elon Musk's "Department of Government Efficiency" on multiple occasions recently. These meetings reportedly focused on a project dubbed "cderGPT," likely named after the FDA's CDER, which oversees drug regulation in the United States.

This development might not be sudden. Wired pointed out that the FDA sponsored a fellowship in 2023 aimed at creating large language models for its internal operations. Furthermore, Robert Califf, FDA commissioner from 2016 to 2017, mentioned that agency review teams have been experimenting with AI for several years.

"It will be interesting to hear the details of which parts of the review were 'AI assisted' and what that means," Califf remarked to Wired. "There has always been a quest to shorten review times and a broad consensus that AI could help."

The FDA has also been exploring the use of AI in other areas of its work.

Califf added, "Final reviews for approval are only one part of a much larger opportunity."

Commissioner Makary, appointed during the Trump administration, has often voiced his enthusiasm for AI technology.

"Why does it take over ten years for a new drug to come to market?" he questioned on X (formerly Twitter) on Wednesday. "Why are we not modernized with AI and other things?"

The FDA's move reflects a wider trend of AI adoption within federal agencies, particularly noted during the Trump administration. For instance, in March, OpenAI unveiled ChatGPT Gov, a version of its chatbot engineered for secure handling of sensitive government data. Elon Musk has also been advocating for the rapid development of another AI chatbot for the US General Services Administration and has explored using AI to attempt a rewrite of the Social Security computer system.

Despite the enthusiasm, the risks associated with using AI in a medical setting are significant. An ex-FDA staffer, who spoke to Wired after testing ChatGPT for clinical applications, highlighted the chatbot's tendency to fabricate plausible-sounding falsehoods. This issue of AI hallucination is a persistent challenge in the industry.

"Who knows how robust the platform will be for these reviewers' tasks," the former FDA employee commented to the magazine.

More on medical AI: Nonverbal Neuralink Patient Is Using Brain Implant and Grok to Generate Replies