In a fireside chat Wednesday afternoon, Food and Drug Administration Commissioner Robert Califf had a lot to say about the future of the agency.

Califf provided an update on the FDA’s hopes for its 2024 budget, the future of AI in healthcare, drug shortages and its ongoing reorganization process.

Here are the main takeaways from the discussion, which was hosted by the Alliance for a Stronger FDA.

FDA’s 2024 budget

Earlier in January, Congress passed a continuing resolution that will keep the FDA funded at least for the next month. 

However, Califf noted lawmakers will need to pass a full year budget in order for the agency to achieve its more ambitious goals.

This isn’t a particularly new or surprising development considering the deeply partisan nature of Capitol Hill but one that Califf is prepared to navigate for the sake of the health agency.

“We have a veteran staff at FDA that has gone through this many times before and people are used to it, but it’s uncomfortable,” Califf said.

Even if Congress does manage to pass the new budget and it doesn’t increase compared to years past, however, that means the FDA will be looking at ways to make cuts.

“Let’s assume Congress comes to its senses and actually passes a budget — if the budget is flat, given our inflationary salary increases, we’ll have to look at a reduction of activities we do because we have a limited amount of people and an overwhelming amount of work,” Califf explained.

The potential of AI in healthcare

Califf spent a significant portion of the discussion focused on AI’s burgeoning role in healthcare. He noted that the FDA has spent plenty of time thinking about the role of regulation in the sphere and what aspects of AI the agency should even regulate.

The FDA already has a “good scheme” to regulate AI algorithms that are embedded in traditional medical devices like a pacemaker or defibrillator, Califf said. Still, when it comes to the incorporation of AI into health systems and clinical practice — such as its use in decision support for physicians — there’s more of a grey area in terms of what the FDA may or may not be subject to regulating.

“Administrative functions are not regulated by FDA,” Califf explained, adding that the agency is only involved when it comes to “patient safety and significant risk.”

A recent study released this week and published in JAMA noted that AI and large language models (LLMs) will quickly impact everyday clinical tasks — like clinical notes, medications and patient data — “with the least scrutiny” from the FDA. The study suggested this could potentially be harmful to patients if action isn’t taken.

Without statutory changes from Congress, however, the FDA does not even have the legal authority to regulate AI in these specific clinical scenarios.

In addition, the agency would have to hire three to four times as many people to “take on” the expected burden of AI regulation in clinical practice, Califf said. That’s especially true for situations and products that use predictive AI, which continues to accrue data and experience and change over time.

“You’re not just approving something like a drug that’s going to be the same drug 10 years later,” Califf said. “With generative AI, that implies something that has a life of its own — something that’s not easily controllable. To deal with this, we need an ecosystem that’s much more accountable than it’s been.”

The FDA is currently in the process of working with external groups to agree upon standards and methods that would validate predictive and generative AI so the algorithms aren’t just initially safe but rather “over time [the AI is] doing what it’s supposed to be doing.”

“We know for sure if we put an AI algorithm into a health system and walk away and leave it, it’s as likely to go in the wrong direction as it is in the right direction,” Califf said. “It has to be continuously tuned.”

When asked about the use of AI in preclinical drug development, such as pharma companies using it to find new targets, Califf opted for a hands-off approach.

“It’s a company’s business how it wants to pick up a target or construct a molecule,” he said. “When it gets into animal and human studies, that’s where the FDA has a primary role and essential responsibility. A naive person may say FDA should be in there regulating how pharma uses AI to figure out a target, but that’s not our job. Our job is to deal with [bringing a product] to development.”

The FDA will likely be impacted by AI internally as well. Similarly to how AI will soon summarize clinician notes and replace much of the human role in electronic health records in doctor’s offices. AI also may assist in making FDA reviewers’ jobs more efficient and address supply chain issues, such as in the case of current drug shortages.

“There is so much progress that could be made with digital supply chains, such as being able to model them [with AI] — like where we’re having these generic drug shortages right now,” Califf said. “I see this as transformative for the internal operations of the FDA.”

Of course, achieving much of this comes down to funding.

“It’s very hard to get Congress to give us capital to make the changes that would be revolutionary,” Califf said. “We have to be ready to do that so we can do our jobs efficiently in a world that’s expanding so much.”

Drug shortages

Califf did not delve into detail on the current drug shortages in the U.S. — with treatments spanning across diabetes, cancer and attention-deficit hyperactivity disorder (ADHD) being impacted — but he briefly suggested that it’s not all the FDA’s fault.

Most of the drugs currently in shortage are cheaper generics, suggesting that the less expensive a product is, the more likely it is to go into shortage. Califf called it an “economic failure.”

“You don’t have to be a business genius to figure out why that’s happening,” Califf said. “There’s a lot we can do at FDA inspections to help deal with the problem,” but ultimately it’s “a system that’s fundamentally broken.”

He continued: “I’ve got to keep the heat on getting agreement in the political sphere to do something about this and fix this fundamental economic problem. We’ve got a real problem — it’s not new but it’s getting worse.”

Food safety reorganization

It’s been a busy month for the agency, which also officially announced its sweeping reorganization plan — the “largest in its history” — just a few weeks ago.

That reorganization effort, which aims to make its inspections more efficient and effective, was largely spurred by the agency being embroiled in controversy over how it handled the infant formula crisis that began in 2022.

Califf said the agency has “made a lot of progress since the infant formula” controversy started, particularly in the infant formula inspection area. Now, the FDA has a dedicated team and “enhanced communications and decision-making” to make sure they can respond quickly, he added.

Health misinformation/COVID-19

COVID-19 has largely fallen to the wayside, transitioning from emergency crisis to endemic disease, four years after the virus took off in the U.S. 

Still, Califf noted that it’s something the FDA is thinking about — particularly when it comes to misinformation.

He was candid in stating it’s still a major problem: “I don’t think we’re winning the misinformation battle,” he said. “Just look at the latest data on vaccination rates.”

In addition, thousands of people are still dying every day in the U.S. from COVID-19, with the Centers for Disease Control and Prevention’s latest data showing weekly deaths ranging from around 1,800 to over 2,000 in January.

“You’re talking about saving tens of thousands of lives that could’ve been saved this [last] year by vaccines,” Califf said.

He acknowledged the persistent sentiment among many Americans who continue to be suspicious of government overreach and emphasized the importance of not only federal agencies — but also industry players — to make scientific information “available in a better way.”