“Once you get a really big jump in performance, the whole paradigm shifts.” - Geoffrey Hinton, pioneer of neural networks
Promise of rapid breakthroughs
The excitement around DeepMind’s AlphaFold in 2020 raised hopes that drug discovery would soon accelerate. Solving the protein folding problem was hailed as a revolution for medicine. Yet the anticipated surge of new treatments has not materialised. Despite massive investments, discovering a new drug still moves at a slow and costly pace.
The paradox of progress
Computing power continues to double while drug discovery productivity keeps falling. This reversal, described as Eroom’s Law, shows that more hypotheses do not translate into better outcomes. For decades researchers have generated millions of ideas about how molecules might influence disease, but only a tiny fraction survive real testing.
[Eroom’s Law observes that the cost of developing new drugs has doubled
roughly every nine years, despite advances in technology. It highlights
declining research productivity in the pharmaceutical industry, driven
by rising regulatory complexity, higher failure rates, and increasing
scientific challenges,opposite to Moore’s Law, hence the name spelled
backward.]
Quantity cannot replace intuition
AI systems can expand the number of possibilities to astonishing scales, but they cannot add intuition or creativity. The real challenge is not generating options but judging their scientific quality. Human insight remains essential because biological systems are messy, unpredictable and resistant to clean patterns.
Why Alphafold was different
AlphaFold thrived because it tackled a well defined question with abundant data and a clear notion of what a correct answer looked like. Drug discovery is nothing like that. It resembles exploration shaped by uncertainty, much closer to spotting hidden talent on a dusty field than solving an exam problem.
The deeply human nature of discovery
History shows that great medicines often emerged from curiosity and serendipity rather than algorithmic logic. AI can support researchers, but it cannot yet imagine breakthroughs. Drug discovery remains a domain where human judgment still guides the frontier.
Summary
Despite AlphaFold’s success, AI struggles in drug discovery because the field demands intuition, creativity and exploration rather than pattern solving. Algorithms can multiply hypotheses but cannot reliably judge their quality in unpredictable biological systems.
Food for thought
If AI cannot yet imagine scientific breakthroughs, how should we rethink the balance between machine power and human insight in medicine?
AI concept to learn: Protein-folding problem and AI
The protein folding problem involves predicting a protein’s
three-dimensional structure from its amino acid sequence, a task long
considered one of biology’s greatest challenges. AI has transformed this
field, with systems like DeepMind’s AlphaFold achieving near-atomic
accuracy by learning patterns from vast structural datasets. These
models help researchers understand diseases, design new drugs, and
engineer novel proteins far faster than traditional laboratory methods,
marking a major breakthrough at the intersection of biology and
artificial intelligence.
[The Billion Hopes Research Team shares the latest AI updates for learning and awareness. Various sources are used. All copyrights acknowledged. This is not a professional, financial, personal or medical advice. Please consult domain experts before making decisions. Feedback welcome!]
COMMENTS