Drug Discovery and AI
Getting a molecule from a lab bench to an approved drug is a pipeline with years-long stages. AI shortens a few of them — and still can't skip a single one of the rest.
“AI discovered a new drug” makes it sound like a model can conjure a medicine out of nothing. What actually happens is narrower and still useful: getting a molecule from a lab bench to an approved drug is a pipeline with years-long stages, and AI meaningfully shortens a few of them — without skipping a single one of the rest.
The pipeline from molecule to approved drug
Before a drug reaches a pharmacy shelf, it passes through distinct stages that together typically take a decade or more, and the vast majority of candidates that enter the pipeline never make it out the other end.
- 1
Target identification
Find the specific biological mechanism a drug would need to act on.
- 2
Candidate discovery
Search an enormous space of possible molecules for ones that might act on that target.
- 3
Lab and animal testing
Test whether the most promising candidates actually behave as predicted outside a computer.
- 4
Human clinical trials
Three phases, each larger than the last, checking safety and then effectiveness in real patients.
- 5
Regulatory approval
A regulator reviews the full trial data before the drug can be prescribed to the public.
Where AI actually shortens it
Candidate discovery is where AI has made the clearest dent. Searching the space of chemically possible molecules by hand or by brute-force lab testing is slow — a model that has learned which molecular shapes tend to bind to a given target can narrow millions of theoretical candidates down to a shortlist worth actually synthesising and testing, in a fraction of the time a purely manual search would take.
Target identification benefits too, in a smaller way — models that scan research literature and biological data can surface plausible targets a human researcher might not have connected on their own.
Where it still can't skip a single step
Nothing about a faster shortlist changes what has to happen after it. A candidate a model is confident about still has to survive lab testing, animal testing, and three separate phases of human trials, because none of those stages exist to find a good molecule — they exist to prove a specific molecule is safe and effective in an actual human body, which a model trained on existing data cannot certify on its own.
Key takeaways
- A drug's path to approval runs through target identification, candidate discovery, lab and animal testing, human trials, and regulatory review — typically a decade or more.
- AI's clearest advantage is narrowing an enormous space of candidate molecules down to a shortlist worth testing, in candidate discovery.
- A promising AI-flagged candidate still has to pass lab testing, animal testing, and three phases of human trials — none of that is optional.
- AI shortens the earliest stage of the pipeline; it does not shorten the years of clinical testing that exist specifically for patient safety.