Precision engineering could be used to return natural immunity to crops that have lost it over time

The open-source AI model improves transparency in predicting how proteins interact with other molecules, which could speed up drug discovery.

Artificial intelligence is transforming drug discovery by analyzing vast amounts of data and improving the efficiency of the traditionally slow and expensive process. In a webinar hosted by The Conversation, experts Jeffrey Skolnick from Georgia Tech and Benjamin P. Brown from Vanderbilt University discussed AI's potential to address key challenges in biomedical research. Traditional drug development has a low success rate, with many candidates failing due to ineffectiveness or harmful side effects. AI can scan petabytes of biological data to identify connections between diseases, predict protein structures and functions, and suggest new therapeutic approaches, including off-label uses for existing drugs. Skolnick highlighted how deep learning models, such as those using transformers and attention mechanisms like in AlphaFold, enable the discovery of higher-order correlations that humans might miss. This includes understanding disease interrelationships and trajectories, potentially leading to broad-spectrum treatments rather than narrow ones for individual conditions. Brown explained different drug modalities, from small molecules to gene therapies, and how AI builds on decades of molecular dynamics simulations. While AI accelerates virtual screening, protein design, and target identification, experts caution that it is not a replacement for experimental validation and clinical trials. The technology offers hope for treating complex diseases like cancer, autoimmune disorders, and metabolic conditions by integrating collective biological knowledge and proposing alternatives when standard treatments fail. Overall, AI promises to make drug development faster and more reliable, though realistic expectations are necessary.