Artificial intelligence (AI) and machine learning (ML) are becoming increasingly important tools in drug discovery, helping scientists manage growing volumes of data and uncover insights that would be difficult to identify through traditional approaches alone. As biotechnology companies seek to accelerate research while maintaining scientific rigor, AI can become a valuable complement to human expertise.
In this Innovation Spotlight, Peter Tummino, president of research and development at Nimbus Therapeutics, discusses how AI is being applied across target discovery and drug development. He covers the challenges of integrating AI into scientific workflows and opportunities that could shape the future of biotechnology innovation.
Peter Tummino, PhD
President of Research and Development
Nimbus Therapeutics
Has the recent rise of AI/ML changed the overall approach to drug discovery and development?
AI and ML are genuinely evolving our work in drug discovery and development. The amount of data and information available to drug developers has exploded over the past 20 years, and the pre-AI approach simply could not keep up with absorbing and integrating large amounts of key information. AI and ML are powerful tools up to the task. Our primary use of AI today is to take in vast amounts of data, summarize it, analyze it, and use the outputs to provide important insights to scientists to guide their decision-making.
How can AI/ML help scientists find the most important targets and identify promising drug candidates?
Finding the most important targets and identifying promising drug candidates are quite different endeavors, and hence, AI plays a distinct role in each. In target discovery specifically, the identification of valuable, novel drug targets is less about traditional data analysis and more about searching, integrating, and analyzing very large databases.
A primary source of information for new target discovery is published literature on basic science, such as peer-reviewed journals. AI can shine here by “reading” and summarizing many publications, sometimes hundreds, in a short period of time. Other important sources include public information about drug discovery activities related to a target, including press releases and patents. AI can also help assess clinical indications by summarizing key considerations, including unmet medical needs, standard of care, and the attributes needed in future therapies.
Today, AI is a powerful tool for providing insights across these areas and informing decisions around novel target discovery. However, Nimbus is not currently using AI to make final decisions about which targets are the most important. In target selection, the scientist still plays the key role, while AI serves as enabling support. Target selection is just one example, but it illustrates a broader point: understanding the science is essential to ensuring that AI is guided by, rather than substituting for, experienced drug developers.
How have Nimbus scientists integrated AI/ML into their existing drug discovery strategies? What sets Nimbus apart in this area?
There are a few major components of AI applications in drug development: the AI algorithms, the data and information sets being analyzed, and the effectiveness of the scientist in applying AI to their specific drug discovery challenges. Nimbus’ perspective is that the first of these components has major similarities across drug discovery groups in the biotechnology industry, while the second is entirely company- and task-specific. Many groups have similar goals and approaches. We believe that today, the distinguishing element is the third component, the willingness and capability of every drug discovery scientist to apply powerful AI approaches to areas that they have specialized in for their entire careers.
The challenge is initially a cultural one, requiring scientists to hand over some of their independence to a powerful assistant, AI. If a drug discovery group does not succeed at this first step, the potential of AI cannot be fully realized.
At Nimbus, we have had a major initiative to integrate AI into our culture of inquiry and discovery. We have reached a point where all scientists are suggesting and trying novel ways to apply AI in their daily workflows. At the same time, we are still early in that journey because much of AI’s potential has not been tapped yet. In general, we start by first defining the problem before selecting the tool. AI is one instrument in a broader scientific toolkit, not a strategy in itself.
What are the key limitations of AI/ML use in drug discovery, and how do your scientists overcome them?
I would point to two areas: data and judgment. AI is only as good as the data being analyzed. Many data sets are flawed and not reproducible, and AI cannot provide valuable insights based on highly flawed data. Therefore, Nimbus and others in the broader biotechnology industry must ensure we are curating the highest-quality data sets. Companies that treat data curation as a foundational capability will have a lasting advantage.
The second limitation is that, today, AI cannot replace the valued judgment of experienced scientists. At Nimbus, we implement discipline when it comes to AI. We stay grounded, set realistic expectations, and keep our scientists at the center of decisions that require genuine creativity and judgment. It is important to deploy AI thoughtfully, by identifying the specific problems it can solve well and those it cannot, and maintaining scientific rigor, which remains essential in drug discovery.
Can you provide an example of how AI tools have improved a specific aspect of your drug discovery process at Nimbus?
One example is our use of AI tools to help scientists navigate the volume of candidate data generated during drug design, and evaluate and prioritize potential molecules more efficiently. At Nimbus, we have used AI to assess proposed molecules against known parameters, such as potency and selectivity, to help scientists determine which drug candidates might be worth pursuing further.
What has been valuable is not that AI replaces human judgment, but that it can help process the most relevant information faster and support more efficient triage. In that sense, AI has improved the speed and efficiency of certain parts of our drug discovery process, while our scientists remain responsible for applying their expertise and interpreting results.
How do you expect the role of AI in biotech to evolve in the next 5-10 years?
Over the next 5-10 years, I expect AI to become a more deeply integrated part of drug discovery, particularly in areas where it can help scientists process large amounts of information and identify patterns. The most immediate gains will likely come from targeted applications where AI can reduce time spent on time-consuming but lower-judgment tasks. The companies that will benefit most from AI will be those that apply it thoughtfully and keep experienced scientists at the center of decision-making.
