Loading
What changed between revision 5 and revision 6, compared over the article's markdown source with its frontmatter — so a change to status, evidence or reversibility shows up as the edit it is.
Older · revision 5
Import
Re-import of content/ai-drug-discovery.md — the file changed after the initial import. Not an edit.
23.4 kB
Revision 5
Revision 6
lines 8–14 → 8–147 unchanged lines not shown
categories: ["foundations", "longevity"]categories: ["foundations", "longevity"]tags: ["machine learning", "drug development", "generative chemistry", "target discovery", "clinical trials", "pharmacology"]tags: ["machine learning", "drug development", "generative chemistry", "target discovery", "clinical trials", "pharmacology"]summary: "The use of machine learning to select drug targets, generate candidate molecules and predict failure, so far demonstrably faster at making compounds than at making medicines."summary: "The use of machine learning to select drug targets, generate candidate molecules and predict failure, so far demonstrably faster at making compounds than at making medicines."updated: "2026-10-28"updated: "2026-10-29"humanEvidence: "AI-derived small molecules have reached phase 1 and 2 trials, with a first phase 3 begun in 2026; sponsors of an mRNA therapy built on algorithm-selected neoantigens report a phase 3 in melanoma met its endpoints, and no drug from either route is approved."humanEvidence: "AI-derived small molecules have reached phase 1 and 2 trials, with a first phase 3 begun in 2026; sponsors of an mRNA therapy built on algorithm-selected neoantigens report a phase 3 in melanoma met its endpoints, and no drug from either route is approved."access: "Platforms are proprietary to pharmaceutical companies or licensed to them, and some models and datasets are public; no molecule designed this way is available outside a trial, though AI-suggested uses of approved drugs are."access: "Platforms are proprietary to pharmaceutical companies or licensed to them, and some models and datasets are public; no molecule designed this way is available outside a trial, though AI-suggested uses of approved drugs are."reversibility: "context"reversibility: "context"lines 50–57 → 50–5735 unchanged lines not shown
language models reading that literature directly. The most consequential public case islanguage models reading that literature directly. The most consequential public case isBenevolentAI's identification of baricitinib, an approved rheumatoid arthritis drug,BenevolentAI's identification of baricitinib, an approved rheumatoid arthritis drug,as a plausible COVID-19 treatment in early 2020.[^richardson2020] Baricitinib went on to beas a plausible COVID-19 treatment in early 2020.[^richardson2020] Baricitinib went on to beauthorized for that use. It is a genuine result and a narrow one: the system reranked knownauthorized for that use. The result was real but narrow: the system reranked known pharmacologypharmacology rather than producing a molecule.and did not produce a molecule. **Molecule generation** trains models over SMILES strings, molecular graphs or three-dimensional**Molecule generation** trains models over SMILES strings, molecular graphs or three-dimensionalposes, then steers them with reinforcement learning or conditions them on a binding pocket.poses, then steers them with reinforcement learning or conditions them on a binding pocket.lines 59–66 → 59–661 unchanged line not shown
claiming designed inhibitors in weeks rather than months.[^zhavoronkov2019]claiming designed inhibitors in weeks rather than months.[^zhavoronkov2019] **Property prediction** estimates absorption, metabolism, cardiac ion-channel liability and toxicity**Property prediction** estimates absorption, metabolism, cardiac ion-channel liability and toxicitybefore synthesis. This is the least glamorous application and probably the most useful, because itbefore synthesis. It is the least glamorous of these tasks and probably the most useful, becausekills bad series cheaply.it kills bad compound series cheaply. **Screening triage** trains a classifier on a modest set of measured compounds and applies it to**Screening triage** trains a classifier on a modest set of measured compounds and applies it tolibraries too large to test. A graph neural network trained on a few thousand molecules testedlibraries too large to test. A graph neural network trained on a few thousand molecules testedlines 81–88 → 81–8814 unchanged lines not shown
coordinating work across contract laboratories. These are where developers of general-purposecoordinating work across contract laboratories. These are where developers of general-purposelanguage and agent models have entered the field, by a different route from Alphabet's, which spunlanguage and agent models have entered the field, by a different route from Alphabet's, which spunits drug-design work out into a separate company. Anthropic reported in October 2026 on researchits drug-design work out into a separate company. Anthropic reported in October 2026 on researchpartnerships with six drug developers — BigHat Biosciences, Bristol Myers Squibb, Divergence Labs,partnerships with six drug developers (BigHat Biosciences, Bristol Myers Squibb, Divergence Labs,Enveda, Generate Biomedicines and The Column Group — and on programmes of its own, having acquiredEnveda, Generate Biomedicines and The Column Group) and on programmes of its own, having acquiredMithril Medicine, a one-person company working on two rare forms of muscular dystrophy whose founderMithril Medicine, a one-person company working on two rare forms of muscular dystrophy whose founderhad used the company's Claude models to coordinate experiments across eleven externalhad used the company's Claude models to coordinate experiments across eleven externallaboratories.[^anthropic2026] Anthropic's stated reasoning is that the bottlenecks are distributedlaboratories.[^anthropic2026] Anthropic's stated reasoning is that the bottlenecks are distributedlines 98–104 → 98–1049 unchanged lines not shown
{ "year": "2020", "title": "First AI-designed molecule in the clinic", "text": "DSP-1181, from Exscientia and Sumitomo Dainippon Pharma, enters a phase 1 trial in Japan for obsessive-compulsive disorder. It was later discontinued." }, { "year": "2020", "title": "First AI-designed molecule in the clinic", "text": "DSP-1181, from Exscientia and Sumitomo Dainippon Pharma, enters a phase 1 trial in Japan for obsessive-compulsive disorder. It was later discontinued." }, { "year": "2020", "title": "Antibiotic found by screening model", "text": "The Collins laboratory identifies halicin using a graph neural network, the first widely cited antibiotic hit attributed to deep learning." }, { "year": "2020", "title": "Antibiotic found by screening model", "text": "The Collins laboratory identifies halicin using a graph neural network, the first widely cited antibiotic hit attributed to deep learning." }, { "year": "2021", "title": "AlphaFold2 released", "text": "Predicted structures for most of the human proteome are made public, changing what structure-based design can be attempted." }, { "year": "2021", "title": "AlphaFold2 released", "text": "Predicted structures for most of the human proteome are made public, changing what structure-based design can be attempted." }, { "year": "2021–2022", "title": "AI-derived target and molecule reach humans", "text": "Insilico's ISM001-055, a TNIK inhibitor for idiopathic pulmonary fibrosis, enters first-in-human testing after both target and compound were nominated computationally." }, { "year": "2021-2022", "title": "AI-derived target and molecule reach humans", "text": "Insilico's ISM001-055, a TNIK inhibitor for idiopathic pulmonary fibrosis, enters first-in-human testing after both target and compound were nominated computationally." }, { "year": "2024", "title": "Consolidation", "text": "Exscientia and Recursion merge, combining two of the largest AI-first drug discovery companies after neither had produced an approved medicine." }, { "year": "2024", "title": "Consolidation", "text": "Exscientia and Recursion merge, combining two of the largest AI-first drug discovery companies after neither had produced an approved medicine." }, { "year": "2026", "title": "First AI-designed small molecule enters phase 3", "text": "Insilico Medicine begins a 320-patient randomized placebo-controlled phase 3 trial of rentosertib in idiopathic pulmonary fibrosis in China, the furthest any generatively designed small molecule has gone." }, { "year": "2026", "title": "First AI-designed small molecule enters phase 3", "text": "Insilico Medicine begins a 320-patient randomized placebo-controlled phase 3 trial of rentosertib in idiopathic pulmonary fibrosis in China, the furthest any generatively designed small molecule has gone." }, { "year": "2026", "title": "Neoantigen therapy meets phase 3 endpoints", "text": "Merck and Moderna report that intismeran autogene, an individualized mRNA therapy encoding up to 34 algorithm-selected neoantigens, met its recurrence-free survival endpoint with pembrolizumab in resected melanoma." } { "year": "2026", "title": "Neoantigen therapy meets phase 3 endpoints", "text": "Merck and Moderna report that intismeran autogene, an individualized mRNA therapy encoding up to 34 algorithm-selected neoantigens, met its recurrence-free survival endpoint with pembrolizumab in resected melanoma." }lines 131–137 → 131–13726 unchanged lines not shown
Merck and Moderna announced that intismeran autogene, an individualized neoantigen mRNA therapyMerck and Moderna announced that intismeran autogene, an individualized neoantigen mRNA therapycovered under [[personalized-mrna-cancer-vaccines]], met its primary endpoint of recurrence-freecovered under [[personalized-mrna-cancer-vaccines]], met its primary endpoint of recurrence-freesurvival and a key secondary endpoint of distant metastasis-free survival when added tosurvival and a key secondary endpoint of distant metastasis-free survival when added topembrolizumab in a 1,137-patient trial in resected stage IIB–IV melanoma. The claim is topline, frompembrolizumab in a 1,137-patient trial in resected stage IIB-IV melanoma. The claim is topline, froman interim analysis, and detailed data had not been released as of late August 2026.[^merck2026]an interim analysis, and detailed data had not been released as of late August 2026.[^merck2026]Each course is built by sequencing the patient's tumour and encoding up to 34 of its mutations,Each course is built by sequencing the patient's tumour and encoding up to 34 of its mutations,selected by what Moderna describes as a proprietary algorithm developed with Merck.[^moderna2023]selected by what Moderna describes as a proprietary algorithm developed with Merck.[^moderna2023]lines 140–157 → 140–1572 unchanged lines not shown
phase 3 depends entirely on where the label's boundary is drawn.phase 3 depends entirely on where the label's boundary is drawn. The newest claims for the approach are not near patients at all. Anthropic's October 2026 reportThe newest claims for the approach are not near patients at all. Anthropic's October 2026 reportmixes molecule-design results — roughly 35,000 antibody designs generated for Divergence Labs inmixes molecule-design results with operational ones. On the design side it lists roughly 35,000about a month, about 1,010 antibody variants made and tested with BigHat Biosciences over about tenantibody designs generated for Divergence Labs in about a month, about 1,010 antibody variants madeweeks, a shortened genetic component of a CAR T-cell therapy that Generate Biomedicines aims toand tested with BigHat Biosciences over about ten weeks, and a shortened genetic component of a CARextend to [[cancer|solid tumours]] — with operational ones, among them a manufacturing-documentT-cell therapy that Generate Biomedicines aims to extend to [[cancer|solid tumours]]. Thereview at Enveda that the company says cut human review time by 89%, and an assessment ofoperational work includes a manufacturing-document review at Enveda that the company says cut humanclinical-trial plans at Bristol Myers Squibb. Every figure is reported by the companies themselves,review time by 89%, and an assessment of clinical-trial plans at Bristol Myers Squibb. Every figureand Anthropic describes the work as early and mostly preclinical: nothing from it has entered ais reported by the companies themselves, and Anthropic describes the work as early and mostlyhuman trial, and the earliest first-in-human date the company offers, late 2028, is conditional onpreclinical: nothing from it has entered a human trial, and the earliest first-in-human date theits candidates continuing to hold up. Its furthest-advanced internal programme, a bispecificcompany offers, late 2028, is conditional on its candidates continuing to hold up. Itsantibody for a form of congenital muscular dystrophy, rests on a pilot study in mice whosefurthest-advanced internal programme, a bispecific antibody for a form of congenital muscularaccompanying technical report states that it was not peer reviewed and that the experiments weredystrophy, rests on a pilot study in mice whose accompanying technical report states that it was notneither randomized nor blinded.peer reviewed and that the experiments were neither randomized nor blinded. > [!caution] What "AI-discovered" means> [!caution] What "AI-discovered" means> No regulator recognizes the category, and companies apply the label to everything from a molecule> No regulator recognizes the category, and companies apply the label to everything from a moleculelines 212–220 → 212–22054 unchanged lines not shown
measured in decades. Compressing discovery from four years to one changes little when the limitingmeasured in decades. Compressing discovery from four years to one changes little when the limitingstep is a trial nobody knows how to design. Better preclinical models, whether [[organoids]] orstep is a trial nobody knows how to design. Better preclinical models, whether [[organoids]] orsimulation approaches such as [[human-digital-twins]], plausibly matter more tosimulation approaches such as [[human-digital-twins]], plausibly matter more to[[healthspan|healthspan research]] than faster chemistry does — though the simulation side is[[healthspan|healthspan research]] than faster chemistry does. The simulation side, however, isleast developed in exactly the systems aging research needs, metabolism and immunity, and noleast developed in exactly the systems aging research needs, metabolism and immunity, and no modelmodel of a whole person exists.of a whole person exists. ## Risks## Risks removed, struck through added, underlinedLine numbers count the serialised markdown of each revision, frontmatter included.
259 lines are the same in both revisions, and 211 of them are not shown. Neither revision here is necessarily the current article — read AI drug discovery for the text this wiki stands behind today.