AstraZeneca is betting on generating artificial intelligence (AI) to advance the development of 30 additional drugs by the end of the decade.
Speaking this week at the AI Summit New York conference, Anna Berg Asber, vice president of research and development for IT at AstraZeneca, told conference attendees that investment in AI will play a critical role in enabling the pharmaceutical company to top $80 billion in annual revenue by the end of the decade.
In fact, AstraZeneca is already leveraging generative AI to streamline research and development by, for example, making it possible to use natural language to query curated medical content using a Research Assistant it created.
In addition, AstraZeneca is also streamlining clinical trial workflows in a way that uses multiple types of AI models to enable researchers to develop new drugs and identify potential new use cases for existing pharmaceuticals, she adds.
In each use case, AstraZeneca is trying to democratize access to knowledge in a way that enables the existing workforce to be more productive versus having to add massive amounts of additional headcount to expand the company’s product portfolio, notes Berg Aber.
As such the user interface through which generative AI becomes more accessible to employees becomes a critical design decision, she adds. “A simple user interface is key,” says Berg Asber.
The most critical thing to remember is to make sure that generative AI is being applied to a problem that is big enough to justify a return on investment (ROI) versus resolving more marginal use cases that can be addressed after initial traction is gained, she adds.
The challenge is making sure that senior leaders buy into those use cases to reduce any friction that might arise as generative AI applications are deployed, notes Berg Asber. Generative AI, for example, isn’t going to replace the need for a radiologist but as roles evolve and change, the support for business leaders is clearly crucial, she adds.
In general, pharmaceutical companies have an advantage when it comes to generative AI because the data they collect is already classified. As a result, aggregating the unstructured data needed to train AI models becomes an easier undertaking than it might be for an organization that would otherwise need to spend more time applying tags to data in a way that an AI model can understand.
Most organizations, however, are clearly focused on applying generative AI to existing workflows versus building completely new applications. Eventually, generative AI will be ubiquitously adopted, but in the short term there are vertical industries that will enjoy a first mover advantage.
Of course, it won’t be too long before generative AI simply becomes the next table-stakes capability that every application is going to be expected to provide. In that context, the race to add generative AI capabilities simply to remain competitive is already on. The next major challenge will be developing an application that provides a truly sustainable competitive advantage for the long haul.
