A recent survey of chemists, carried out by Elsevier’s Reaxys team, showed that only 13% identified a familiarity with machine learning (or similar technologies) as being important for future chemists.
Perhaps that number makes some sense, when you think of chemistry as the kind of hard science that consists of working out calculations and producing reactions in labs. But in the 21st century, when the most advanced digital technology drives innovation, this is really alarming news.
More than that, though, it is dangerous thinking from a business perspective. Today’s innovation-hungry companies cannot afford to put deep learning, which is a kind of artificial intelligence, on the back-burner. Why? Because deep learning is poised to become one of the biggest drivers of innovation in the chemistry industry.
Deep learning is centered on machines learning and extrapolating based on data sets, and it has the power to help chemists do some of their most time-consuming, often repetitive work in a far shorter period of time. With machines assisting with the creation, synthesis and application of compounds and molecules, think of how much faster researchers will be able to identify the best possible solutions.
Though machines can help us make amazing strides with their ability to process so much information so quickly, they are not creating those data sets—they are dealing with what is given to them. That means that developing the deep learning technology isn’t the only challenge – finding “clean,” reliable data to power the machine’s work is also vital, difficult work.
In this recent article in R&D Magazine, I discuss the challenges and opportunities that deep learning presents in greater detail.
All opinions shared in this post are the author’s own.