Built To Avoid Risk: How Industry Norms Stall Drug Discovery

At Discovery & Development US 2026 in San Diego, discovery leaders named the industry norms they would change.

Drug discovery runs on new ideas, but many of its biggest barriers aren't scientific. They're industry habits, collectively adopted norms. At Discovery & Development US 2026 in San Diego, Drug Discovery Online asked researchers and executives one question: What industry norm would you change?

Their answers point to the same problem. Practices that once reduced risk now slow progress.

Starting From The Platform, Not The Biology

Danielle Dettling, CSO and co-founder of Fury Biosciences, previously led CAR-T and CD3 bispecific programs at Pfizer. She says biologics companies too often build around their own tools. "Most biologics companies are focused on their particular platform/modality, and then they take that tool and try to apply it," she said. "I would like to see them really focus on the biology, and then figure out how to best direct a particular platform or modality."

Learning Too Late

Pablo Tamayo, Ph.D., professor of genomics and precision medicine at UC San Diego and co-developer of Gene Set Enrichment Analysis, wants that biological context understood "before we move to clinical trials." He says more than 90% of oncology trials fail because "we don't do a good job in understanding if something will generalize or what is the domain where it will work."

Anil Tarachandani, Ph.D., has held translational medicine roles at Verge Genomics, Pfizer, Biogen and Merck. He sees the same delay in how trials are sequenced. "We go through a long process of clinical trials, collect phase one data, and then go to the phase two data," he said. "Wouldn't it be nice if you could collect lots of the endpoints in phase one that enable us to de-risk the phase two?"

The "You First" Mentality

James J. Hickman, Ph.D., chief scientist at Hesperos, has seen his lab's human-on-a-chip efficacy data used in a Sanofi IND. He wants the field to move from animal models "towards these new alternative methods which are proven to work better than the animal models at this point."

Adoption is slow because few want to go first. "I think the hesitancy to try new things, the 'you first' mentality," said Chris Worrall, Ph.D., VP of U.S. business development at Nanoform. "They wait for someone else to go through and do it, and that can hinder not only novel approaches but novel drugs getting to market."

AI faces the same resistance. Petrina Kamya, Ph.D., VP and global head of AI platforms at Insilico Medicine, said that "you either have people that are all in or people that are super skeptical." Parmi Shah, CEO of Rasayan Labs and co-inventor on a patent for turning legacy lab notebooks into AI-ready drug discovery data, says the gap is widest outside big pharma. "The top 10 big pharma companies are really monopolizing the AI market," she said. "The mid-size pharma companies and biotech labs still have an opportunity to use AI in a big way. They just need to leverage startups."

Cost As An Afterthought

Craig Behnke, Ph.D., EVP of production and development at Lumen Bioscience, co-authored the first GPCR crystal structure. He says companies treat patient cost as an afterthought. "That's how you end up with very, very expensive drugs that aren't accessible to most of the world's population," he said. At Lumen, "it's on the top line of the [Target Product Profile] TPP at the very beginning of the project."

The Common Thread

Each of these norms can be seen as cautionary behavior about cost, risk, or precedent. Together they keep the industry stuck with the inefficiencies it wants to get rid of.

Yet, we have all conformed to these norms over our careers. That’s just how we work, we adapt.

I’m grateful to all the fantastic delegates who shared their opinions on a somewhat sensitive topic. Also grateful for those who are not just adapting but truly championing innovative execution and blasting past the norms.

Finally, I’m grateful to my colleague, Tom von Gunden, chief editor of Drug Delivery Leader, for graciously recording the video interviews and breaking the norms with me in San Diego.

To all our readers, what norms would you change?

Full Transcript (edited for clarity)

[0:15] Danielle Dettling: There are a bunch of them, but I just want to comment on one. I work primarily in biologics.

Most biologics companies are focused on their particular platform or modality, and then they take that tool and try to apply it. Rather than starting from that angle, I would like to see them really focus on the biology and then figure out how best to direct a particular platform or modality.

[0:50] Craig Behnke: I would like to change where the final cost of the drug to the consumer or the patient ends up on the target product profile. For a lot of people, it's a total afterthought, and that's how you end up with very, very expensive drugs that aren't accessible to most of the world's population. For us, it's on the top line of the TPP at the very beginning of the project.

I think it's something that's traditional in our industry. It's not a concern, because the way the payer system works, in the United States in particular, takes that off of everybody's mind. Also, the general cost of developing a drug is so high that it's hard to think about that at the very beginning.

But if you can somehow bring that in at the beginning of your discovery project and your development project, and then go into clinical development, we believe you can really make drugs more accessible to a much larger share of the world's population.

[1:50] Chris Worrall: Ooh, that's a good question. For me, I think it's the hesitancy to try new things, the "you first" mentality. I think there are a lot of novel approaches and novel technologies out there that a lot of people are reluctant to try because maybe the path hasn't been blazed already.

And so they wait for someone else to go through and do it, and that can hinder not only novel approaches but novel drugs getting to market.

If they were a little bit less risk-averse, then perhaps they might be able to get life-changing medicines through to patients more quickly, and potentially even more of them.

[2:45] Parmi Shah: I think what's happening currently with AI is that the top 10 big pharma companies are really monopolizing the AI market. The problem with that is that a lot of mid-size pharma companies and small biotechs are still doing the work traditionally, in wet labs.

And instead of leveraging AI tools, they're mostly using foundation models to get to their answers. Maybe they should try task-specific models and look for companies offering them at a much lower cost, rather than the in silicos of the world, who are gatekeeping the price at $30,000 a seat.

So I think mid-size pharma companies and biotech labs still have an opportunity to use AI in a big way. They just need to leverage startups.

[3:30] James J. Hickman: I would love to change the use of animals in drug discovery and move more toward these new alternative methods, which are proven to work better than the animal models at this point.

Drugs that utilize animal data only are about 10% effective. You can't get much worse.

[4:05] Anil Tarachandani: I would say it's in clinical trials. What's interesting to me is that we go through a long process of clinical trials, collect phase 1 data, and then go to phase 2.

Wouldn't it be nice if you could collect many of the endpoints in phase 1 that would enable us to de-risk phase 2? That's one thing I would like to change.

[4:25] Graham Marsh: One change I would love to see within the industry is more openness to data sharing. In a space where data is so valuable, being more willing to collaborate, share, and contribute to large data sets and collaborative efforts is something that would drive the field forward faster and more efficiently, and it would benefit everybody.

So how do we get people to share data? I don't know, but I'd love to see that happen.

[4:55] Adamo Sulpizi: Yeah. Talking about discovery and development, of course, I think the collaboration between preclinical and clinical teams should be far more focused and connected.

In fact, advancements in the preclinical phases are then moved into the clinical phases in order to give the patient the right product.

[5:15] Petrina Kamya: I would say the extremes of people.

I'm going to be controversial. You either have people who are all in or people who are super skeptical. I'd like to see more people who are in between, more accepting of AI for its true capabilities, and willing to adopt it.

[5:35] Ryan Kole: I'd like to make it easier to get drugs into clinical trials.

I think it's a long process to do all the preclinical work needed. I'd also love to figure out ways to not have to do it in animals first, and to get drugs that could help a lot of people into trials sooner, while keeping them safe, of course.

[6:05] Pablo Tamayo: Okay. That would be to find the right context. When we develop a biomarker, a target, or a patient population that we plan to target with a therapeutic, we should find out much better what the underlying context is, so that when we look at something in model systems, preclinically, we understand it better before we move to clinical trials.

It seems to me that more than 90% of clinical trials in cancer fail because we don't do a good job of understanding whether something will generalize or what the domain is where it will work. I think that rule would go a long way to improve things, in my opinion.

[7:05] Shubhankar Dutta: So, one industry norm I would change.

There are so many things I wish I could change, but one thing I would like to change is this: Whenever someone presents an idea, be open to it and think it through. Don't discard it with, "Oh, it's not possible." Don't discourage. Always encourage: "Okay, let's look into it." Think through the pros and cons, and give it some thought.

If no one has thought of it in the past, that doesn't mean it's not going to work. Science always comes out of deep thinking. If someone is thinking logically, give the logic a chance: Does it make sense for the experiment or not? That's what I want to change. Think it through.