FDA Wants an AI Tool Called ‘Elsa’ to Review Drug Safety

There was once a time when the US Food and Drug Administration (FDA) scrutinized the safety and efficacy of the drugs and medical devices created by massive companies with dreams of making billions off of your illness. Scientists, with years of experience and expertise informing every decision, did as well as a human possibly could at protecting the American populace from anything ranging from common mistakes to dangerous snake oil.

Under the Trump administration, the thoughts and experiences of those human beings have been outsourced to Elsa. Elsa is an AI large language model running on Amazon’s GovCloud, a platform built for U.S. intelligence agencies.

In a recent Journal of the American Medical Association op-ed, FDA officials Marty Makary and Vinay Prasad pitched a five-point plan to “unleash AI” on the agency’s pile of paperwork. The FDA’s ambitions for Elsa include reviewing 500,000-page drug applications, summarizing clinical data, generating regulatory code, and greenlighting novel food dyes for kids.

Prasad, it should be noted, was a proponent of COVID-era public health mandates.

FDA Wants to Use AI to Approve Drugs Faster. Experts Aren’t Sure That’s a Good Idea.

The federal agency that protects us against the horrors of the medical industry’s insatiable greed is outsourcing a ton of its work to an AI chatbot.

As covered by Futurism’s Joe Wilkins, critics are skeptical, to say the least. Dr. Reshma Ramachandran of Yale accused the FDA of parroting Big Pharma’s wishlist, arguing that the agency’s AI push reads “straight out of PhRMA’s playbook.”

While Makary and Prasad insist they’ll avoid getting too cozy with the drug industry, the government’s recent track record doesn’t exactly inspire confidence.

Remember: we’re talking about an FDA that’s already tossed out 2,000 staff members. Many of them now seemingly replaced with the technology that could be the future of human civilization, or it could be the tech flavor of the day before the industry shifts to the new hot thing, whatever that may be.

None of this is encouraging, especially after RFK Jr.’s “Make America Healthy Again” report was, possibly, almost definitely slopped together by AI. It was found to have cited studies that don’t exist, a common fault of generative AI.

It’s unclear whether that report was created by Elsa, though it wouldn’t be surprising.