AI in drug discovery is a revolution in progress, but guardrails are needed

Lorikeet News Desk

published

February 11, 2025

Credits: Anna Shvets via Pexels

Key Points

AI is increasingly integral to pharmaceutical drug discovery, but security risks remain a concern.
Sergey Jakimov and Garri Zmudze, managing partners at LongeVC weigh in on where they see the greatest opportunities.

AI is a toolkit to make something happen. It's not a product in its own right, but a toolkit to accelerate certain processes and help the company deliver on the end result.

Sergey Jakimov

Managing Partner | LongeVC

AI is now deeply interwoven with pharmaceutical drug discovery. While AI is making significant leaps in finding new solutions for life-changing medicines, ensuring the process is safeguarded with security standards remains paramount.

AI for drug discovery ramping up: AI is becoming synonymous with drug discovery but experts warn it’s not always understood:

"There are many AI for drug discovery companies, but when we look at those projects, having spent many years within [the industry] sometimes they seem a bit naïve," says Garri Zmudze, managing partner at LongeVC. "There are also strong big players, and from my perspective, that's the future of new drug development. It's already here."

LongeVC is an international venture capital firm investing in exceptional early-stage founders with a mission to bring breakthrough biotech to market. By focusing on longevity and biomedical advancements, LongeVC seeks to back companies that not only innovate but also demonstrate industry traction:

"With AI and drug discovery, you look at two things: whether they have gained traction with the industry and whether they have their own pipeline. You don't look at companies that just sell drug discovery services to Big Pharma. That doesn't work and has proven to be a faulty strategy. If you have something unique, you will have your own assets and want to develop them yourself," added Sergey Jakimov, also a managing partner at LongeVC.

Initial success: Within LongeVC’s portfolio is one of the pioneers in AI-driven drug discovery, Insilico Medicine. The company currently has two drugs in Phase 2B trials, positioning itself as a leader in the age-related and longevity sector.

"Insilico is the first to enter phase two clinical trials with an AI-created asset. This is the first molecule to pass into human efficacy studies, with good initial data already announced. It's a pretty big deal," noted Zmudze.

Any application of machine learning or AI is based on whether it makes sense and contributes to the end goal the company is pursuing.

Sergey Jakimov

Managing Partner | LongeVC

Addressing the risks: While AI-driven drug discovery presents transformative opportunities, it also introduces significant security risks. With AI models relying on vast datasets, often containing sensitive biomedical data, privacy needs to be safeguarded. Strong validation procedures are necessary to ensure the AI’s reliability, and securing against malicious exploitation of the research is vital.

FDA guardrails: Regulatory oversight remains top of mind, especially with a new administration making changes across government organizations. In regards to how it relates to drug development, Jakimov explained, "There is only one regulatory aspect we can really look at, and that's the Food and Drug Administration guidelines in the US and the European agency guidelines in Europe. The way it works for us is that whenever you look at the company... you can only invest if they have a specific disease indication. Only through having a specific disease indication can you gather an IND enabling data package, get the ID approval, and then get into the human phase to make a drug out of your clinical program. It's the FDA language that the companies need to be able to speak."

AI as a tool, not a standalone product: "Any application of machine learning or AI is based on whether it makes sense and contributes to the end goal the company is pursuing. AI is a tool to make something happen. It's not a product in its own right, but a toolkit to accelerate certain processes and help the company deliver on the end result," emphasized Jakimov.

One more thing