AI is regarded by some top executives at big pharma (GSK and others) as a tool to uncover not only new molecules, but also new targets. Ability of deep neural networks to build ontologies from multimodal data (e.g. “omics” data) is believed to be among the most disruptive areas for AI in drug discovery, alongside with data mining from unstructured data, like text (using natural language processing, NLP).
There is a considerable trend for “AI democratization” where various machine learning/deep learning technologies become available in pre-trained, pre-configured “of-the-shelf” formats, or in relatively ready-to-use formats -- via cloud-based models, frameworks, and drag-and-drop AI-pipeline building platforms (for example, KNIME). This is among key factors in the acceleration of AI adoption by the pharmaceutical organizations -- where a non-AI experts can potentially use fairly advanced data analytics tools for their research.
Proof-of-concept projects keep yielding successful results -- in research studies, and in the commercial partnerships alike. For example, companies like Recursion Pharmaceuticals and Exscientia achieved important research milestones using their AI-based drug design platforms.