How A.I. Is Disrupting Pharma: The High-Stakes Race to Revolutionize Drug Discovery

As the FDA prepares to regulate A.I. in drug development, biotech startups and pharma giants are racing to harness its potential for faster, more precise drug discovery.

Illustration of a small person surrounded by oversized hands and floating pills, symbolizing the impact and control of artificial intelligence and pharmaceutical companies over drug development and patient outcomes.
The rise of A.I. in the pharmaceutical industry has placed unprecedented power in the hands of tech-driven drug discovery, raising questions about innovation, control and patient safety. Unsplash+

For better or worse, A.I. in the pharmaceutical industry has reached a fever pitch—a point of no return. Ever cautious, the Food and Drug Administration (FDA) has stepped in to regulate the burgeoning use of A.I. in drug discovery and development, vowing to “promote innovation and protect patient safety.” But while the agency ponders policy, companies like Immunai have already charged ahead. The biotech company recently inked an $18 million deal with AstraZeneca, which aims to harness Immunai’s A.I.-powered immune system model to optimize clinical trials. Startups like Insilico Medicine and Recursion Pharmaceuticals tout A.I. as their secret weapon in the hunt for new drugs, though skeptics whisper that their claims smack of A.I. washing.

Sign Up For Our Daily Newsletter

By clicking submit, you agree to our <a href="http://observermedia.com/terms">terms of service</a> and acknowledge we may use your information to send you emails, product samples, and promotions on this website and other properties. You can opt out anytime.

See all of our newsletters

“There is a spectrum of opinions on whether A.I. is going to solve problems on its own in drug development, versus whether it’s not going to be relevant and feasible at all. I think I sit somewhere in the middle,” Raviv Pryluk, co-founder and CEO of PhaseV, tells Observer. Pryluk, who previously served as SVP of operations and analytics at Immunai, struck out on his own in 2023 to launch a company focused on using A.I. and machine learning to streamline clinical trials. “It can bring a tremendous amount of value, but only when done carefully,” he cautions.

The FDA claims that since 1995, it has received over 300 submissions for drugs and biological products containing A.I. components. Yet the complexity—and promise—of A.I. has grown exponentially, says Dave Latshaw II, formerly the head of A.I. drug development at Johnson & Johnson. “It’s a bit different these days, primarily through the scale and quality of data that we have access to,” Latshaw tells Observer.

In practice, A.I. has already reshaped clinical trials, from pairing patients with studies to targeting specific populations. But there’s a catch: while narrowing down patient types might yield more accurate results, it also restricts the commercial appeal of a drug. “They want to find the exact type of person their drug will be useful for but, at the same time, doing that limits the scope of the total population it will be useful for and limits the total commercial impact,” Pryluk says.

Latshaw, who left Johnson & Johnson in 2020 to co-found BioPhy, sees an ongoing tension between technological advancement and commercial priorities. He launched BioPhy to address the inefficiencies he couldn’t touch within J&J’s bureaucratic walls. For him, the real promise of A.I. is in refining the entire drug discovery pipeline, including clinical trial optimization.

Mergers and acquisitions also present a tantalizing opportunity. Pharmaceutical giants often rely on acquisitions to bolster their portfolios, and A.I. can sift through troves of pre-clinical and clinical data to spotlight the most promising drugs. “Machine learning and A.I. can help run over the pre-clinical and clinical data and all the available literature in order to make better decisions on whether to acquire this drug versus that drug,” Pryluk notes.

A.I. could also help diversify clinical studies, a longstanding challenge in the industry. Since 1993, the National Institutes of Health (NIH) has mandated the inclusion of women in research, but gender gaps—particularly in preclinical animal studies—remain a stubborn problem. Researchers could tailor studies using advanced methods like causal machine learning to ensure that findings apply across demographics. “What if we can use more sophisticated and granular methods like causal machine learning, for example, and ask causal questions on whether the results that we’ve seen on subsets of women patients actually can tell us something, and then we can expand the population?” Pryluk asks. “I think this type of method can bring more women, more people from minorities, to participate in clinical trials.”

Pryluk believes that precision medicine—treating each patient as a unique case—will be the industry’s next seismic shift.

Of course, there’s still plenty of skepticism. Every A.I. application in pharma has to survive a gauntlet of risk-benefit analysis. Data sharing, cybersecurity, inscrutable algorithms and biases pose perennial challenges. For instance, genetic data—the lifeblood of many A.I. models—raises significant privacy concerns. Just look at the ongoing debates around platforms like 23andMe.

Steven Aviv, CTO of Pentavere, sees data trust as paramount. Pentavere’s DARWEN A.I. sifts through unstructured healthcare data to identify patients for specific treatments. “Pharmaceutical customers tell us that trust in the data for these applications is imperative,” Aviv tells Observer. That’s why Pentavere leans on Databricks’ Data Intelligence Platform, he says. The secure framework is designed to handle massive data volumes without compromising compliance.

Latshaw, too, emphasizes transparency in A.I. solutions. Interpretability is key, he says, noting that researchers risk misinterpreting A.I.’s outputs without it. “There’s always a motivation for publishing the next even moderately state-of-the-art method for something, but sometimes you have to ask if the question you’re addressing in the research is actually the right question,” Latshaw cautions. “Just because you can predict the structure of the protein doesn’t mean you can accurately predict what its function is going to be and how that impacts the disease.”

Pryluk is blunt about the stakes. “We are not in a game. We are talking about patients. People are dying and it needs to be practical,” he says.

Despite the inevitable setbacks, Latshaw is bullish on A.I.’s long-term impact. “There’s going to be failures in the space as there always are when people push the envelope,” he says. “But there’s going to be an incredible amount of learning from that, and the byproduct is going to be higher productivity in discovering new molecules.”

As the FDA debates regulation, it’s inviting input from industry stakeholders. Latshaw reminds us that the agency’s goal is patient protection, not industry obstruction. “If they see the industry wanting to use new technology, it’s their responsibility to make sure it’s used in a way that doesn’t impact that overall goal,” he says. “The strength of that regulation will depend largely on how much of an impact the technology will have on the end recipient.”

In Latshaw’s view, the payoff could be transformative: over the next 5–10 years, he expects a boom in viable drug candidates reaching clinical trials. “What that means for the companies is they’re going to be able to bring a lot more drugs to market, but also get that many more drugs to the patients that need them,” he says.

How A.I. Is Disrupting Pharma: The High-Stakes Race to Revolutionize Drug Discovery