The FDA is using Generative AI to drastically cut drug evaluation times

Disease is everywhere. Drug discovery is getting insanely fast to meet the challenge, and as a result the Food and Drug Administration (FDA) is having to try to get new Artificial Intelligence (AI) developed drugs through testing and approved as fast as possible – with the goal eventually being real time. As a consequence the organisation is trying all kinds of new tools to help it move faster and they’ve just announced that they’ll be launching Generative AI across all of their centers by July.

The announcement follows the appointment of the agency’s first chief AI officer, Jeremy Walsh, and the completion of an AI-assisted scientific review pilot.

The generative AI rollout will allow FDA staff to spend less time on “tedious, repetitive tasks,” ultimately making the review process more efficient, according to the agency.

“I was blown away by the success of our first AI-assisted scientific review pilot. We need to value our scientists’ time and reduce the amount of non-productive busywork that has historically consumed much of the review process. The agency-wide deployment of these capabilities holds tremendous promise in accelerating the review time for new therapies,” said FDA Commissioner Dr. Martin Makary.

The technology offers the possibility of dramatically shortening the review and drug approval process - an analysis on MDO’s sibling site Drug Discovery & Development estimates that it could save  hundreds of thousands of review hours annually at the FDA’s Center for Drug Evaluation and Research (CDER).

“This is a game-changer technology that has enabled me to perform scientific review tasks in minutes that used to take three days,” said Jinzhong (Jin) Liu, deputy director, Office of Drug Evaluation Sciences, Office of New Drugs in CDER, in the press release announcing the rollout.

This technological push for efficiency comes along with massive staff reductions. The FDA, among many other agencies within the Department of Health and Human Services (HHS), has been the target of Elon Musk’s Department of Government Efficiency’s (DOGE) mass layoffs. Many senior leaders at the FDA were swept up in the plan to cut the HHS workforce by 20,000 people.

Details on the AI platform are sparse, but Wired reports that the FDA has been meeting with OpenAI, one of the biggest names in the AI space. While no contract has been signed, sources close to the project say people with OpenAI, the FDA and DOGE are talking about initiatives including cderGPT and Research GPT, according to Wired.

While the prospect of a shortened review process and bringing new drugs to market faster is appealing, the “aggressive timeline” the FDA has set for embracing generative AI is being met with some criticism and concern.

“It’s as if the FDA is so enamoured with freeing its scientists from ‘non-productive busywork’ that it forgot to mention how it will prevent the productive work from going off the rails,” Markus Brinsa, founder, president and CEO of strategy consulting firm SEIKOURI wrote on LinkedIn.

The agency plans to have all of its centers operating on a unified generative AI platform integrated with its internal data platforms by June 30.  FDA processes necessarily involve sensitive and proprietary data. What data is being used to train the platform? How is data being stored and secured?

An HHS spokesperson told Medical Design & Outsourcing that the FDA’s use of generative AI is meant to support, rather than replace, human experts. But there are still questions about the balance between technology and those human experts.

The “human-in-the-loop” mantra keeps people at the heart of decision making; AI makes processes faster and easier, but people are always there during the development and application of these systems. But a recent report from STAT found that doctors don’t always catch the mistakes that AI makes. And there are reports that AI hallucinations — its propensity to give incorrect information or to simply make things up — are getting worse as it gets more powerful.

How does the FDA plan to manage these risks with its rapid deployment of the technology? That remains to be seen.

Chief AI Officer Walsh and Sridhar Mantha, director of CDER’s Office of Strategic Programs, are leading the rollout of generative AI across the FDA. The agency plans to expand its generative AI capabilities as time goes on.