November2024SeizingtheAIforScienceopportunityConorGriffin|DonWallace|JuanMateos-Garcia|HannaSchieve|PushmeetKohliAnewgoldenageofdiscovery2AcknowledgementsThankyoutoLouisaBartolo,ZoëBrammerandNickSwansonforresearchsupport,andtothefollowingindividualswhosharedinsightswithusthroughinterviewsand/orfeedbackonthedraft.Allviews,andanymistakes,belongsolelytotheauthors.ŽigaAvsec,NicklasLundblad,JohnJumper,MattClifford,BenSouthwood,CraigDonner,JoëlleBarral,TomZahavy,BeenKim,SebastianNowozin,MattClancy,MatejBalog,JenniferBeroshi,NitarshanRajkumar,BrendanTracey,YannisAssael,MassimilianoCiaramita,MichaelWebb,AgnieszkaGrabska-Barwinska,AlessandroPau,TomLue,AgataLaydon,AnnaKoivuniemi,AbhishekNagaraj,HarryLaw,TomWestgarth,GuyWard-Jackson,AriannaManzini,StefanoBianchini,SameerVelankar,AnkurVora,SébastienKrier,JoelZLeibo,ElisaLaiH.Wong,BenJohnson,DavidOsimo,AndreaHuber,DipanjanDas,EkinDogusCubuk,JacklynnStott,KelvinGuu,KiranVodrahalli,SanilJain,TrieuTrinh,RebecaSantamaria-Fernandez,RemiLam,VictorMartin,NeelNanda,NenadTomasev,ObumEkeke,UchechiOkereke,FrancescaPietra,RishabhAgarwal,PeterBattaglia,AnilDoshi,YianYin.Introduction34PartA:TheopportunitiesPartB:TheingredientsPartC:TherisksPartD:ThepolicyresponseIntroductionAquietrevolutionisbrewinginlabsaroundtheworld,wherescientists’useofAIisgrowingexponentially.Oneinthreepostdocsnowuselargelanguagemodelstohelpcarryoutliteraturereviews,coding,andediting.InOctober,thecreatorsofourAlphaFold2system,DemisHassabisandJohnJumperbecameNobelLaureatesinChemistryforusingAItopredictthestructureofproteins,alongsidethescientistDavidBaker,forhisworktodesignnewproteins.Societywillsoonstarttofeelthesebenefitsmoredirectly,withdrugsandmaterialsdesignedwiththehelpofAIcurrentlymakingtheirwaythroughdevelopment.Inthisessay,wetakeatourofhowAIistransformingscientificdisciplinesfromgenomicstocomputersciencetoweatherforecasting.SomescientistsaretrainingtheirownAImodels,whileothersarefine-tuningexistingAImodels,orusingthesemodels’predictionstoacceleratetheirresearch.ScientistsareusingAIasascientificinstrumenttohelptackleimportantproblems,suchasdesigningproteinsthatbindmoretightlytodiseasetargets,butarealsograduallytransforminghowscienceitselfispractised.Thereisagrowingimperativebehindscientists’embraceofAI.Inrecentdecades,scientistshavecontinuedtodeliverconsequentialadvances,fromCovid-19vaccinestorenewableenergy.Butittakesaneverlargernumberofresearcherstomakethesebreakthroughs,andtotransformthemintodownstreamapplications.Asaresult,eventhoughthescientificworkforcehasgrownsignificantlyoverthepasthalf-century,risingmorethansevenfoldintheUSalone,thesocietalprogressthatwewouldexpecttofollow,hasslowed.Forinstance,muchoftheworldhaswitnessedasustainedslowdowninproductivitygrowththatisunderminingthequalityofpublicservices.Progresstowardsthe2030SustainableDevelopmentGoals,whichcapturethebiggestchallengesinhealth,theenvironment,andbeyond,isstalling.Inparticular,scientistslookingtomakebreakthroughstodayincreasinglyrunintochallengesrelatingtoscaleandcomplexity,fromtheever-growingliteraturebasetheyneedtomaster,totheincreasinglycomplexexperimentstheywanttorun.Moderndeeplearningmethodsareparticularlywell-suitedtothesescaleandcomplexitychallengesandcancompressthetimethatfuturescientificprogresswouldotherwiserequire.Forinstance,instructuralbiology,asinglex-raycrystallographyexperime...
