22.10.2026 | Published by bit.bio
bit.bio featured in Drug Discovery & Development, Pharma Press
In a new feature for Life Science Daily, bit.bio CEO Dr Emma Pepperell addresses the limits of artificial intelligence in drug discovery. While machine learning excels at high-throughput pattern recognition and process optimisation, Dr Pepperell emphasises that AI cannot compensate for an underlying deficit in mechanistic understanding or infer novel biological mechanisms without wet-lab experimental data.
Because machine learning algorithms remain strictly bound to the quality and validity of their input data, training them on traditional animal models merely reproduces species-specific biases and limitations. To truly reduce late-stage clinical attrition, the industry must commit equally to generating dense, reproducible, human-relevant datasets, such as those produced using induced pluripotent stem cells (iPSCs), to ground computational models in human biological reality.
"The true ceiling of AI-driven drug discovery will not be determined by computational scale alone, but by our capacity to generate the dense, high-fidelity, and human-relevant biological datasets required to ground these models in clinical reality."
Emma Pepperell CEO, bit.bio