The FDA is continuing its push to adopt emerging artificial intelligence tools, this time unveiling an agentic AI program that employees can choose to use in their work.
The agency intends for staff to use AI agents for complicated tasks like meeting management, premarket reviews, post-market surveillance and inspections, the FDA said in a Dec. 1 release.
In agentic AI, models are designed to generate outputs in furtherance of a specific goal. Different parts of the model complete different steps in the process, with outputs ultimately integrated through a decision-making process with human oversight.
“We are diligently expanding our use of AI to put the best possible tools in the hands of our reviewers, scientists and investigators,” FDA Commissioner Marty Makary, M.D., said in the release. “There has never been a better moment in agency history to modernize with tools that can radically improve our ability to accelerate more cures and meaningful treatments.”
The new AI model has not been trained on data submitted to the FDA, the regulator said, and is hosted on a high-security GovCloud environment.
Makary announced in May that he intended for the FDA to fully integrate generative AI into its work by the end of June; that pronouncement was followed by the early debut of Elsa, a large language model meant to speed up the agency’s review times.
Elsa was accused of being rushed after its rollout, with reports that the model was inaccurate when asked to name FDA leaders or summarize information about FDA-approved products.
In the Dec. 1 release, the FDA claimed that 70% of its staff have voluntarily used Elsa.
With the FDA's workforce significantly reduced by large-scale layoffs earlier this year, including on the regulator’s review teams, the agency has been searching for ways to hasten its workflow as delays continue to occur. The FDA has missed multiple target decision dates for potential drug approvals this year, most recently pushing back the PDUFA date for Denali Therapeutics’ Hunter syndrome therapy by three months.
In addition to its AI efforts, the FDA recently claimed it was kick-starting a hiring spree to add more reviewers to its roster, but no jobs have yet been listed on the U.S. government's official job board.
The agency has also launched an unprecedented new “national priority” voucher program meant to offer recipients reviews in just one or two months. This program has since drawn scrutiny from lawmakers and been publicly lambasted by George Tidmarsh, M.D., Ph.D., the recently ousted head of the FDA’s Center for Drug Evaluation and Research.
Nov. 21, hundreds of biopharma leaders signed a letter to Makary expressing worries about the FDA’s ability to function, citing a survey in which 82% of respondents raised concerns about the regulator’s leadership turnover, missed deadlines, extensive layoffs and its “failure to stand up for established vaccine science and important mRNA research,” among other concerns.