As artificial intelligence (AI) enhances an expanding range of applications, from ChatGPT to clinical decision-making and far beyond, bioprocessing will not be left behind. The bioprocessing industry’s interest in AI, though, is not new. Semantic Scholar, itself an AI-driven platform, shows a steady increase in related articles since the 1950s.

The application of AI in bioprocessing drugs, however, only started to kick in during the past decade. According to scientists from Jiangnan University in China: “Recently, artificial intelligence  emerged as a data-driven technique independent of the complex interactions used in mathematical models and has been gradually applied in bioprocess[ing].”

In fact, the Jiangnan scientists showed that various forms of AI can improve many aspects of bioprocessing, including modeling and process optimization. Nonetheless, they noted: “There are some potential challenges and research gaps which may limit its further application in bioprocess[ing].”

Crucial challenges include collecting the necessary data and making use of it. As these scientists pointed out, “a growing number of [bioprocessing] studies have been conducted, however, it is quite difficult to integrate their data for further analysis because experimental conditions in [the research] vary widely.”

Some scientists express concerns

Even some scientists at large pharmaceutical companies express concerns over AI’s current application to bioprocessing. As one example, Christos Varsakelis, PhD, associate director of artificial intelligence/machine learning at the Janssen Pharmaceutical Companies of Johnson & Johnson, and his colleagues indicated a wide use of AI in bioprocessing R&D, but most of those projects land “in limbo in the so-called Proof of Concept purgatory.” Like the scientists at Jiangnan University, Varsakelis and his colleagues indicated an ongoing data challenge, because “no company has a sufficiently diversified data or product portfolio to proceed on its own.”

Instead of trying to revolutionize all of bioprocessing in one go, AI might whittle its way into one application at a time. For instance, a group of scientists at Sartorius Stedim Biotech used machine vision combined with convolutional neural networks to monitor a bioprocess for foam production, which is common and can reduce a culture’s productivity and even create blockages in a bioreactor. Even when using just a smartphone camera in the Sartorius project, the scientists concluded that this method shows “great promise for the application of machine vision to implement cheap, flexible, and robust monitoring for foam control for upstream bioprocessing.”

So, will AI transform bioprocessing? One day, it probably will, but not yet—at least not in a way that disrupts the entire spectrum of the industry.