Prioritize a drug candidate
Useful candidate discovery; laboratory validation and clinical development remain substantial parts of the work.
SCIENCE UNDER AUTOMATION
AI in Drug discovery: capabilities and evidence
Experimental validation of a nominated compound
What has AI demonstrated?
Experimental validation of a nominated compound
- Test
- Halicin identified by a model and tested for antibacterial activity.
- Comparison
- Comparative discovery yield and matched human-screening baseline not extracted.
- System and version
- Stokes et al. · Cell 2020
- Evidence checked
- 2026-10-20
A validated candidate is not a clinical drug. This record remains a case study, not a field-level productivity estimate.
What does the field require?
Nominate and test a therapeutic candidate
- Rank promising molecules
Halicin demonstrates computational nomination followed by experimental testing.
Stokes et al. · Cell 2020 - Run and interpret biological assays
Experimental follow-up uses human laboratory work.
Stokes et al. · Cell 2020 - Validate safety and mechanism
A candidate discovery does not establish general autonomous validation.
Stokes et al. · Cell 2020
Advance a treatment through clinical testing
- Design an informative clinical study
One AI-origin candidate does not evaluate autonomous trial design.
Rentosertib phase 2a trial - Recruit and run a trial
The reported clinical trial is conducted by human clinical teams.
Rentosertib phase 2a trial - Assess efficacy and safety
Clinical interpretation remains human; phase 2a is not a discovery-system comparison.
Rentosertib phase 2a trial
Research tasks
Can AI replace the whole workflow?
The results above measure specific tasks. The field assessment also requires completion of the connected workflow, including the requirements listed here. A field-wide replacement percentage has not been established.
Read the assessment method