In some ways, machine learning is nothing new. The term itself dates back to 1958 when the computer scientist Arthur Lee Samuel coined it. “The pharmaceutical industry has been using data science, machine learning, and AI in some form for at least 25 years, if not longer,” said Kailash Swarna, a managing director and Accenture Life Sciences Global Research and Clinical lead. But something has changed with the technology shifting from niche to mainstream. New research from Accenture reveals that 87% of biopharmaceutical R&D leaders now view AI and machine learning as crucial to their success.
“What’s different this time is the unprecedented pace at which these new technologies have come to the industry,” Swarna emphasized. “It has really changed the way people think about AI, making it more accessible to everybody.” A recent Accenture report, “Reinventing R&D in the age of AI,” highlights both the immense potential and significant challenges of integrating AI and machine learning into pharmaceutical R&D.
From niche to necessity
According to the Accenture report, companies that successfully wield AI could potentially bring new medicines to market four years faster than otherwise while cutting development costs by 35-45%.
That doesn’t mean that biopharma companies need to shoot for the stars. Even a 1% improvement in clinical success rates could translate to hundreds of millions in saved revenue. But while this potential is significant, implementing AI is not a simple plug-and-play solution by itself. “It’s not like a traditional CIO role focused just on IT; it’s about the power of technology to transform various business segments in the organization.” This new reality demands a new breed of talent—one fluent in both the language of science and the language of AI.
The rise of the “bilingual” scientist
“We’re seeing the emergence of what we call ‘bilinguals’ – people who speak both the domain language and the technology language,” Swarna notes. This shift is reshaping the skills required in the industry, demanding a new breed of professionals who can bridge the divide between science and technology. “It’s no longer enough to be just a medicinal chemist; you need to understand computational science and chemistry, including concepts like language models for molecules,” Swarna explains. On the flip side, computer scientists venturing into biopharma must grasp the complexities of molecular biology and how their algorithms can be applied to solve real-world problems in drug discovery.
Reimagining education for an AI-powered future
This convergence is already revamping education. Universities are no longer confining disciplines to silos, instead forging new paths with interdisciplinary degrees like computational chemistry, bioinformatics, and cheminformatics. These programs equip graduates with the fluency needed to thrive in an AI-powered R&D environment. This evolution in talent and skills underscores the profound changes AI is bringing to pharmaceutical R&D, promising to accelerate innovation while demanding new forms of expertise.
This convergence is already being reflected in academia, with universities such as Stanford University and the University of Cambridge breaking down traditional silos to offer interdisciplinary degrees like computational chemistry and bioinformatics. For instance, Stanford’s Biomedical Informatics program integrates computer science, biology, and medicine, while Cambridge offers a Master’s in Bioinformatics and Computational Biology. These programs, often blending elements of computer science, biology, chemistry, and mathematics, are producing a new generation of scientists equipped for the complexities of AI-driven drug discovery. Additionally, institutions like the Massachusetts Institute of Technology (MIT) are offering specialized courses in machine learning for drug discovery and development, directly addressing the intersection of AI and pharmaceutical research. This evolution in talent and skills underscores the profound changes AI is bringing to pharmaceutical R&D, promising to accelerate innovation while demanding new forms of expertise.
This transformation is already being felt on the front lines of research, where AI is changing not just what scientists do, but how they work.
Empowering scientists, transforming R&D
“These tech investments provide a more powerful set of tools that allow our scientists to be more productive and do things better than we’ve been able to do before,” Swarna adds. This shift enables researchers to focus on higher-level problem-solving and interpretation rather than routine tasks.
Swarna envisions the positive implications: “It’s a good problem to have if our pipeline productivity increases by 40%. I’d welcome that because we could then use our scientists in a more strategic way to prioritize and improve productivity.”
This transformation is already reshaping roles within R&D. For instance, traditional biology roles have evolved into separate biology and data science positions, with scientists now spending about a third of their time managing and integrating data with their lab experiments.
The power of language, scaled up
In addition, the integration of genAI is not only changing the way scientists work but also how they interact with technology. “One of the strengths of language models is that they allow us to interact in a natural way,” explains an expert in the field. “An analogy I often use when teaching is that we’re all walking around with a language model in our head, trained on a lifetime of experience with both structured and unstructured input. What we’ve done with this technology is to build it at a human population scale.”
Navigating the challenges, unlocking the potential
While so-called hallucinations continue to be a problem with large language models, they can be mitigated with methods such as retrieval augmented generation (RAG). By grounding the models’ responses in verified data, this new paradigm could enable researchers to ask complex questions and receive answers in a way that was previously unimaginable. “We can now ask a simple, plain language question that’s just a concept in our mind and get results in a way that was never possible before,” the expert adds. “It’s scaling human intellect in an unprecedented manner. We do best when we see results that give us the additional substrate to think about and ask the next set of questions. That’s the real power of what we see happening with the creation of these models.”
Filed Under: Drug Discovery and Development, machine learning and AI