D-Wave Quantum Inc. has teamed up with the pharmaceutical division of Japan Tobacco Inc. to develop a groundbreaking artificial intelligence model leveraging quantum computing to advance drug discovery. The collaboration focuses on building a proof-of-concept model that enhances the training of AI models, specifically generative pretrained transformers, similar to the engines behind large language models like OpenAI’s ChatGPT, but aimed at generating drug molecules instead of words.
The aim is to utilize D-Wave’s quantum processing units (QPUs) to enhance training beyond classical computing’s capabilities. The quantum approach has shown potential in producing a greater number of valid drug molecules than those created via traditional graphics processing units, which are typically used in AI model training. This advancement also allows the model to identify potential drug candidates more effectively, improving upon the dataset trained through conventional methods.
Quantum computing’s ability to navigate the vast chemical space for drug discovery effectively is pivotal, as slight changes in a molecule’s properties can dramatically affect its potential characteristics, such as toxicity and efficacy. Japan Tobacco sees this initiative as a significant step towards faster and more cost-effective discovery of small-molecule compounds. The project signifies the integration of D-Wave’s quantum annealing capabilities in optimizing AI training processes, marking a promising move towards a future where Quantum AI drives drug discovery.
