Google

November2025

AIandtheFutureofLearning

BenGomes

ChiefTechnologist

Learning&Sustainability

ChristopherPhillips

VP&GMEducation

LilaIbrahim

ChiefOperatingOfficerGoogleDeepMind

JamesManyika

SVP

Research,Labs,Technology&Society

YossiMatias

VP&GM

GoogleResearch

01

Abstract

02

LearningMatters

03

Opportunities

AI’sPromiseforLearning

04

Challenges

ImportantAreastoAddress

05

OurWorkonAIinLearning

06

Conclusion

Abstract

AIopensupnewpossibilitiesforpeopletolearn,accesseducation,andfollowtheircuriosity.Itcanalsohelpaddresscriticalchallenges,includingconcerningtrendsinlearningoutcomes,immensepressureoneducators,andalackof

accesstoqualityeducationinmanycommunities.

Googlebelievesthatbysupportingteachersandpersonalizinglearning,AIcanhelpunlockhumanpotentialallaroundtheworld.

Whilehumanrelationshipswillalwaysremaincentraltoteachingandlearning,weseetremendousopportunityforAItosupport

Google’smissiontomaketheworld’sinformationuniversally

accessibleanduseful.Atthesametime,AIpresentsurgent

challengesandunknownsthatsocietymustcollectivelyreckon

with.OurgoalistohelpimprovelearningoutcomesbydevelopingAIproductsthataregroundedincorelearningscienceandin

closepartnershipwiththeeducationcommunity—becausethegreatestpotentialofAIishelpingeveryonereachtheirs.

GoogleAIandtheFutureofLearning3

LearningMatters

Learningisthebedrockofhumanpotentialandsocietalprogress.Itequipsuswithnewskills,sparkscuriosityandtheexchangeoffreshideas,spursinnovativebreakthroughs,anddriveseconomicgrowth.

Newtechnologieshavealwaysbeenaforcetoexpandsociety’s

knowledge,aswellasaccelerate,challenge,andultimatelyreshapehowwelearnandwhatweknow.Fromtheinventionoftheprintingpresstothevastexpanseoftheinternet,advancesintechnologyhavehelped

todemocratizeaccesstoinformationandopportunity,pushingthe

boundariesofhumanpotential.Learning,whilelifelong,isespeciallyrapidinearlychildhood—bothwithinandoutsideofformaleducationsettings.

Curiosityandthedesiretolearnarehumannature,andtoday,learningopportunitiesareaccessibletomorepeoplethaneverbefore.

GoogleAIandtheFutureofLearning4

Learningmatters

GoogleAIandtheFutureofLearning5

Abriefhistoryofeducation:

Wherewestarted,whereweare

Historically,formaleducationwasexclusive:limiting

thepathtoopportunitiestoaselectfew.Asaresult,manyeffortshavefocusedonwideningaccess,withalongarcofprogressfromroyaltutors(effective,

personalized,butinaccessible),totheindustrial

modelofeducation(apowerfultoolforexpanding

accessandliteracy,butnotpersonalized),torecenteducationtechnologiesthathaveallowedteacherstoreachlearnersbeyondtheirclassrooms.

Theworldhasmadehugestridesinaccess.

90%of

primaryschool-agedchildrengloballyareenrolled

inschool

andapproximately

87%completetheir

primaryeducation

.However,challengesremain,

includingunevenaccesstoeducation,differencesineducationquality,limitedcontentavailabilityacrosslanguages,aswellaslowandunequallearning

outcomes.

Globalpopulationofyoungpeopleaged0-19

PeryoungpersonresourcesrelativetoUSA

<1%1-5%5-10%10-30%

>30%

Bucketedbyresourcesavailableperyoungperson,relativetoUSA

????}????

Mostoftheworldhasless

than10%ofUSAresources(GDP)

peryoungpersonwithwhichto

educatetheirpopulation.

Resourcesdefinedas2023GDPatPPP,fromWorldBank;youngpersondefinedasaged0-19,fromUN.GreycircleshavenoGDPdataavailable.

Learningmatters

GoogleAIandtheFutureofLearning6

Makinggoodonthepromiseofeducationmeansensuringalllearnersreachproficiencywhile

supportingtheirpotentialandwell-beingateverylevel.Publiceducationsystemsfacesignificant

strainasaresultofinadequatefunding,the

continuingimpactof

pandemic-eralearningloss

,disruptionfromglobalconflicts,teacherturnoverandshortages,studentattendanceandmental

healthchallenges,andpersistentsocio-economicdisparities.

Theseinterconnectedfactorsarecompoundedbytheneedtoconstantlyupdatecurriculatopreparestudentsforthedemandsofachangingworld.

AccordingtotheOECD’sProgrammefor

InternationalStudentAssessment(PISA),global

learningoutcomeshavedeclinedoverthelasttwo

decades.Resultsfromthe

2022PISAsurvey

,whichcaptureddatafrom81countriesandeconomies,

notedan“unprecedentedperformancedrop”withmeanperformanceinmathematicsdownby15

pointsandreadingdownby10pointscomparedto

2018.Challengesarebeingseenindevelopedand

developingcountriesalike.Anddespitepositive

trendsinenrollment,

272millionchildrenremain

outofschool

(11%ofchildreninprimary,15%in

lowersecondary,31%inuppersecondary).Asthe

schoolagepopulationoflow-andmiddle-income

communitiescontinuestogrow,expertsestimatetheworldwillneed

44millionmoreteachersby2030

toprovideuniversalprimaryandsecondaryeducation,includinginplacesthatalreadyfacemeaningful

barrierstofullysupportingtheirexistingstudents.

Itisnoteasytokeepupwiththeaccelerating

demandsofarapidlychangingworld.Today’sjobs

demandnotjustfoundationalskills,butadvanced

problemsolving,collaboration,andtheabilityto

learnthroughoutlife.TheemergenceofAImarks

anotherperiodofprofoundtransformation—the

opportunitytorethinkhoweducationshouldadapttoboththeopportunitiesandthechallengesahead.AseducationalsystemsgrapplewiththequestionofwhatitmeanstobewelleducatedinaworldwhereAIisubiquitous,studentswillneedanewsetofskillstoprepareforthefuture.

Trendsinmathematics,readingandscienceperformance

PISAtestscores,OECDaverage

MathReadingScience

Source:

OECD(2023),PISA2022Results(VolumeI):TheStateof

LearningandEquityinEducation.

Learning,teachingandeducationareuniversalinhumansocieties

Learning:Theacquisitionandretentionofinformation,representingashiftinwhatanindividualknowsorcando.

Teaching:Theprocessofhelpingsomeoneelselearn.This

happensinbothformaleducationandoutsideofthatprocess.

Education:Thebundlingofteachingandlearninginastructuredsettingalongsidearelativelyconsistentcurriculum.Thisoften

endswithaformalrecognitionofcompletion.Forsimplicity,

weusethetermteachertoindicateaneducatorinthistypeofformalsetting.

Thoughfocusedonprimaryandsecondaryeducation,thispaper'sdiscussionofopportunities,limitations,andfutureexplorationisrelevanttoallages.

Note:AIisabroadtermencompassingawiderangeoftechnologies,includingpredictivemodels,generativeAIrootedinlargelanguagemodels,andgeneralintelligence.Throughoutthispaper,weusethetermAItotalkaboutthecategoryasawhole.

Opportunities:

AI’sPromiseforLearning

AdvancesinAI,suchaslargelanguagemodels,presentnew

opportunitiestoequiplearnerswiththeskillstheyneedforsuccess.

Whiletheinternetremovedbarrierstoaccessinginformation,AI

increasesourabilitytounderstandandapplythatinformation.Thisleapfrompassiveconsumptiontoactive,deepunderstandingisaprofoundchangethatisrevolutionaryforeducation.AIcanexpandteachingandlearningbymodifyingcontenttobemoreengagingandinteractive,

simplifyingcomplextopics,andenablingdeeplypersonalizedlearningatscale.

GoogleAIandtheFutureofLearning7

Opportunities:AI’sPromiseforLearning

GoogleAIandtheFutureofLearning8

Educationsystemshavealwaysfacedacorechallenge:limitedresources.

Ideally,everystudentwouldspendampletime

workingintheir

zoneofproximaldevelopment

–thesweetspotof“justright”learningchallengesthatleadtonewskillgrowth.Butbecausestudentneedsvarysowidely,traditionaleducationcannotalwaysprovidethispersonalizedsupporttoeveryone.

WhileAIisbynomeansperfect,itdoeshavethe

potentialtoreducebarriersandallowpeopletolearnmoreeffectivelythanbefore.Forstudents,AIcan

helpcreateindividualizedlearningpathsandoffer

extrahelp,whetherornottheyhavethesupport

ofaneducator.Foreducators,AIcanserveasa

teachingassistantthatsupportstheirworkloadandenablesnewapproaches,ultimatelyfreeingupmoretimefortheessentialhumanaspectsofteaching.ItiscriticalthatAI-poweredlearningtoolsaredevelopedandsupervisedbyexpertsandeducatorstoensuretheyreflectlearners’needsandprovideeffective,

evenlydistributedpedagogicalbenefits.

Technologistshavemadesimilarpromisesabout

accessandpersonalizationduringpriorwaves

ofeducationtechnology—andwhilesomewere

met,manywerenot.Thesecyclesdid,however,

offervaluablelessonsaboutwhatworks(and

whatdoesn’t)ineducationtechnology,including

thattechnologyalonecannotsolvedeep-rooted

challenges,thatitmustbedesignedwithandforthestudentsandteachersitintendstoreach,andthat

usersmustbeequippedwiththeknowledgearoundhowtoimplementnewtoolsinmeaningfulways.

AIisfarmorepowerfulthanpastinnovations,offeringtrulytransformativepossibilities.However,turning

AI’spromiseintoreliable,everydaypracticewillrequirebothtimeandintentionaleffort.

"WhileAIisbynomeansperfect,itdoeshavethepotentialtoreducebarriersandallowpeopletolearnmoreeffectivelythanbefore."

Opportunities:AI’sPromiseforLearning

GoogleAIandtheFutureofLearning9

lab_research

Unlocking

thepower

oflearning

science

HowAIcantransformlearning

interactive_space

book_ribbon

door_open

cheer

Personalizing

Make(almost)

Removing

Ahelping

teachingand

tutoring

anything

learnable

barriersto

learning

handforbusy

educators

01

Unlockingthepoweroflearningscience

AIpromisestobringtheverybestofwhatweknowabouthowpeoplelearn(learningscience)intoeverydayteaching,andmayevenidentifynewfrontiersforeducationresearch.

Whendesignedandimplementedcorrectly,

AIsystemscanembodytheprovenprinciples

oflearningscience:includingsparkingactive

participation,encouragingdeeppracticeandreview,andweavinginthebenefitsofspacedrepetition.Thiscanenhancelearningeffectivenessandefficiency,leadingtomorebreakthroughmomentsformore

students.AIsystemsthatareexplicitlyfinetunedforthesepurposescancreatenew,scalablelearning

activitiesandassessmentsthatwerepreviously

impossible.Foreducators,thesetoolscanlighten

theburdenofimplementation,makingpowerful

teachingstrategiesmoreaccessibleandachievableinlesstime.Forstudentswhochoosetoengage,

AImovesthemfrompassiveconsumptiontoactive

application—becomingacatalystforcuriosity,more

meaningfullearning,andassessmentsthatfeellesslikememorizationtestsandmorelikeopportunitiestodemonstrateknowledge.

Thepotentialgoesevenfurther:AIcanhelpusdeepenourunderstandingoflearningitself.

Byanalyzingpatternsofstudentengagement,

motivation,and“aha”moments,wecanuncovernewinsightsintohowcuriositygrows,howpersistence

takesroot,andhowlearningflourishesacross

differentcontexts.Perhaps

learningalongsideAI

mayrevealentirelynewfacetsofskilldevelopmentand

humangrowthpotential—remindingusthatlearningisnotjustaboutknowledgetransfer,butabout

expandingtheboundariesofwhatispossible.

Opportunities:AI’sPromiseforLearning

GoogleAIandtheFutureofLearning10

lab_research

Unlocking

thepower

oflearning

science

interactive_space

book_ribbon

door_open

cheer

Personalizing

Make(almost)

Removing

Ahelping

teachingand

tutoring

anything

learnable

barriersto

learning

handforbusy

educators

02

Personalizingteachingandtutoring

Thegoalofpersonalizationisn’tnew,butAIopensupnewwaystobringittolifeatscale.Whatwasonceanaspirationmaynow

bepossible.

Whilestructurededucationhashadsuccessin

teachinglargegroups,itoftenstrugglestomeet

thediverseneedsofindividualstudents.AImay

ultimatelyalloweverylearnertotakeatruly

individualizedlearningjourney,realizingthepromiseofpersonalizationlikeneverbefore.

AIcangiveeducatorstoolsthatcanadaptlessoncontenttoastudent’sspecificneedsandgoals—providingtailoredfeedbackandidentifying

knowledgegapsthathinderlayeredlearning

progress.Itcanalsoadjustbasedonastudent’scontext,likewhetherateacherispresent,what

thesubjectis,andwhathasbeenlearnedbefore.

Educatorscanreceiveinsightsintostudentprogressandlearninggaps,enablingmorepersonalizedandfocusedsupportwhilereducingcognitiveload.Thiscanhelptargetlearnersattherightlevelto

close

understandinggaps

andhelpensurethatlearningdifferences(includingneurodivergence)donot

determineone'spotentialforsuccess.

Outsidetheclassroom,lackoffoundational

knowledge,studyskills,motivation,ormeanscanallpresentbarriersforstudyingorcompleting

homework;ascansimplyfindingsomeoneto

helpyouwhenyou’restuck.AIcanserveasan

inexpensive,non-judgemental,always-available

tutor.Thereisrigorousevidencethat“high-dosage”personalhumantutoringhasoneofthelargest

positiveimpactsonstudentachievement,with

consistentlysignificanteffectsonstudentsfrom

familieswithlowincomes1,2,3,4.WhileAItutorscannotreplacethehumanelementofgreatteachingor

tutoring,theycanbeacomplementorbridgewhenhumansupportisunavailable.AsAItutorsbecomemorewidelyused,itwillbeimportanttounderstandhowmuchofthebenefitfromtutoringcomesfromtheopportunitytoworkwithintheproximalzoneofdevelopmentwithaknowledgeable(potentiallyAI)coachversusthepowerofhumanconnection(suchasthenaturaldesiretoconnectwithsomeonewhocaresaboutyou).

Opportunities:AI’sPromiseforLearning

GoogleAIandtheFutureofLearning11

lab_research

Unlocking

thepower

oflearning

science

interactive_space

book_ribbon

door_open

cheer

Personalizing

Make(almost)

Removing

Ahelping

teachingand

tutoring

anything

learnable

barriersto

learning

handforbusy

educators

03

Make(almost)anythinglearnable

Weliveinanageofinformationabundance.Inorderto

beunderstandable,informationneedstobeavailableandapproachable.

Peoplearemostcuriousaboutthingsthatthey

feelarelearnable,orthattheyare

ontheverge

ofunderstanding

.AIcanstripawayjargonand

adjustexplanationsanddeliverymodestoa

learner’sbackgroundandpreferences—helpingstudentsspendmoretimeintheirzoneofproximaldevelopment.

Thecoreschoolcurriculumiswellcoveredwith

expertlyproducedcontent,butqualityoptionsfalloffrapidlywhenmovingintomorespecializedsubjects.AIopensupnewopportunitiesforpeopletolearn

andretrainthroughouttheirlives,especiallyfornicheinterestsandchangingjobmarketdemands.

Whetherit’sconnectingacademicconceptstoreal-worldapplicationsorbridgingthegapbetweennewinformationandpriorknowledge,AItoolscanhelpmakelearningmoreexciting–fosteringintrinsic

motivationandsparkingcuriosityandexploration.

Fromlefttoright,toptobottom:E

xploretheMuséed’Orsay,Paris

withGoogleAI

,

Woolaroo

,

GoogleLens

,

Movingpaintings

.

Opportunities:AI’sPromiseforLearning

GoogleAIandtheFutureofLearning12

lab_research

Unlocking

thepower

oflearning

science

interactive_space

book_ribbon

door_open

cheer

Personalizing

Make(almost)

Removing

Ahelping

teachingand

tutoring

anything

learnable

barriersto

learning

handforbusy

educators

04

Removingbarrierstolearning

Everyonedeservesaccesstoqualitylearningexperiencesandtheskillsandtoolsneededtobuildthefuturetheyenvisionforthemselves.

Forthemajorityofhumanhistory,accessto

knowledgeandinformationwasacriticalbarriertolearning.Whiletheprintingpressandinternethaveimprovedaccessinrecenthistory,informationis

stillnotuniversallyavailable.Someofthisremains

aninfrastructurechallenge,butbarriersexistevenforthosewithinternetaccess:ascarcityofcontentandlearningopportunitiesinone’snativelanguage,contentformatlimitations,andcontentunsuitable

fordifferentlearninglevels.AIhasthepowertohelpinalloftheserespectsbytransformingcontentintonew,moreaccessibleforms,anddeliveringcontentattherightleveltocloselearninggaps.

Millionsoflearnersstilllackaccesstoformal

educationorsufficienttimewithateacher,orfacechallengesfromphysicalorlearningdisabilities.

AIhasthepotentialtodeliverqualitylearning

experiences

—makingtheworld’sinformationmoresuitabletoindividuals’needs.

AImaynotonlyraisetheceilingforthosewhohaveeducationalsupporttoday,itwillalsosignificantlyraisethefloorforlearnersacrosstheglobe.

GoogleAIandtheFutureofLearning13

lab_research

Unlocking

thepower

oflearning

science

interactive_space

book_ribbon

door_open

cheer

Personalizing

Make(almost)

Removing

Ahelping

teachingand

tutoring

anything

learnable

barriersto

learning

handforbusy

educators

05

Ahelpinghandforbusyeducators

AIcanbecomeapowerfulteachingassistant,freeingupprecioustimeforteacherstofocusonmeaningfulhumanandsocial

interactionwith

students

.

AIcanassistbusyteachersacrossarangeofareas,spanningfromdeepresearchandcontentcreation,toactivityplanningandadministrativetasks.

Separately,AIcanalsoassistbyenablingnewformsofinstructionorassessment.ThiscanincludeAI-ledgroupbreakoutsthatdriveactivestudentlearning,adaptivereal-timefeedbackformultiplestudents

atonce,oropen-endeddebatesthatevaluateargumentsandpushthefrontierofastudent’sunderstanding.

Thisassistancecanbothhelpstudentsfillgapsintheirlearningandenableteacherstospendmoretimewithstudents.Mostimportantly,AIcanhelpteachersgetbacktodoingthejobtheysignedupfor,

focusingontheimportanthumanaspectsof

teaching

:helpingtheirstudentslearnhowtolearn,inspiringtheirinterests,fosteringconnections,

anddevelopingthemascuriousandwell-roundedindividuals.

Challenges:

ImportantAreastoAddress

Likeallmajortechnologicalshifts,AIintroducesnewchallenges.Somearefamiliar,suchasprotectingprivacyandpreventingexposureto

inappropriatecontent,whileothersarespecifictoAI’scapabilities.Whilethesechallengesvarytremendouslyacrossglobalsettings,andmanifestdifferentlyacrossprimary,secondaryandhighereducationlevels,theymustbeconfrontedhead-on.Questionsofthisnature-andmagnitude-arebestansweredtogether,andhelpusmakeprogressasasociety.

GoogleAIandtheFutureofLearning14

GoogleAIandtheFutureofLearning15

Challenges:ImportantAreastoAddress

target

shield

network_intelligence

psychology_alt

crowdsource

Accuracyand

objectivity

Safety

Criticalthinking

Cheatingand

learningloss

Equalaccess

01

Accuracyandobjectivity

AprimaryconcernisthatAImodelscan"hallucinate"andproducefalseormisleadinginformation,similarto

humanconfabulation

.

Whilehallucinationrateshavefallensharplyinthe

pasttwoyearsasmodelsaretrainedtousetrusted

sources

and

verifytheoutputs

,aharderchallengeremains:determiningwhichsourcesaretrustworthyandhowtohandlesubjectivequestions.Thisis

particularlychallenginggiventhebiasesinglobaldata.

AIalsoneedstherighttoneandmustavoid

confirmingincorrectassertionsandproviding

excessivepraise.Itshouldchallengeastudent’s

misconceptionsandcorrectinaccuratestatements,notsimplyactasanuncriticalmirrorthatreinforces

them.Thisisimportantbothinensuringaccuracy,butalsoinmaintainingeducationalmotivationandself-regulation.

ThisisanareaofdeepimportancetoGoogle,and

whilenosystemisperfect,wehavemademeaningfulprogresswitheachnewmodelrelease,andcontinuedrivingtheindustryforwardwithnew

benchmarks

.By

carefullyandresponsiblybuildinganddeploying

AImodels

,weaimtomanagetheserisksaswehaveforourotherbusinesses,likeSearch,sinceour

earliestdays.

Challenges:ImportantAreastoAddress

GoogleAIandtheFutureofLearning16

target

shield

network_intelligence

psychology_alt

crowdsource

Accuracyand

objectivity

Safety

Criticalthinking

Cheatingand

learningloss

Equalaccess

02

Safety

Safetyisamajorconcernforusersofallages,butthereisaheightenedresponsibilitytoprotectyoungerusers.Keyrisksfocusoncontentorinteractionswhichmayposephysicalorpsychologicalharms.

Alayeredsafetystrategy,includingage-appropriateAIliteracyandcontentfiltersisessential,butwill

needtoevolveasfastasthetechnologydoes.

Google’scommitmenttosafetyisunderpinnedby

extensivetestingandcontinuousexpertconsultation.Weconductrigoroususerandadversarialtesting,

including“RedTeaming”exercisesandchildsafety-focusedsprints,specificallydesignedtoidentify

vulnerabilitiesandbuildstrongsafeguardsagainstharmfulcontent.Ourcomprehensivepoliciesand

safeguardsweredevelopedbyGoogle’sdedicatedin-housespecialists,withcontinuousguidancefromthird-partychilddevelopmentexperts.

Educationalinstitutionstaskedwithkeepingstudentinformationsecuremustalsobegooddatastewards–forexample,byomittingpersonallyidentifiable

informationfrompromptswhenusingAItohelpbuildapersonalizedstudentplan.Googlehasimplementedspecificprivacyprotectionstailoredtotheunder-18demographic,recognizingtheirvulnerability

andhigherpropensitytosharehighlypersonalinformation.

Challenges:ImportantAreastoAddress

GoogleAIandtheFutureofLearning17

target

shield

network_intelligence

psychology_alt

crowdsource

Accuracyand

objectivity

Safety

Criticalthinking

Cheatingand

learningloss

Equalaccess

03

Criticalthinking

Acommon

discussion

centersonwhetherstudentsmight“offload”toomuchthinkingtoAI,leadingto“

metacognitivelaziness

,”andcausingthemtolosetheabilitytoengageindeeplearning.

Therealopportunity,therefore,istodesignAIthatpromotes—notreplaces—deepthinking,suchasthroughquestion-ledtutoringthatprompts

reflectionsorasksforexplanations.

AseducationalpsychologistJohnSweller’s

Cognitive

LoadTheory

remindsus,thegoalisnottomaximizestruggle,it’stofocuseffortonthementalworkthatmatters.Unproductivestruggle,likenostruggleat

all,isn’talwaysgoodforlearning.Insomesituations,AIcanhelpbyreducingunproductiveloads,such

asthe"

split-attention

"and"

modality

"effects,bymakingsenseoffragmentedoroverwhelmingtextandimages.Astudent’senergycantheninsteadbechanneledintohigher-orderreasoningand

problem-solving.

Thekeyquestionbecomes:HowdoweensureAItoolsareusedtoexpandcuriosity,creativity,andanalysis,ratherthanshortcutthem?Muchoftheanswerliesinhumanrelationships.Motivation

topursuedifficultthinkingoftendependsontheencouragement,challenge,andsocialdynamics

ofteachersandpeers—somethingnoAIcanfullyreplicate.

Withintentionaldesignanduse,AIcouldhelpfree

moreteachertimefortheseessentialrelationships,

whilealsoscaffoldinglearnerstoengageincomplexreasoningontheirown.AtGoogle,theaspirationistobuildtoolsthatnotonlysupportteachersinthisway,butalsogivelearnersagencyintheirlearning,whileguidingthemintolearningjourneysthatdemand

perseverance,reflection,andcriticalthought.

Challenges:ImportantAreastoAddress

GoogleAIandtheFutureofLearning18

target

shield

network_intelligence

psychology_alt

crowdsource

Accuracyand

objectivity

Safety

Criticalthinking

Cheatingand

learningloss

Equalaccess

04

Cheatingandlearningloss

SurveyssuggestthatsomestudentsuseAIinwaysthatmany

wouldconsidertobecheating5,6.However,thereisalsolittle

consensusonwhereexactlytodrawthelinebetweeneffective

tooluseandcheating7,8,notleastbecauseeducatorshavedifferingviewsonwhereAIuseshouldbeencouragedversusprohibited

andbecausetheappropriateuseofthetechnologyisstillbeingdetermined.

Asaresult,thereare

mixedperspectives

onwhethertheoverallratesofcheatingarehigherthaninthe

pre-LLMera.Aswithallnewtoolsitwilltaketimefornewnormstobeestablishedinresponsetothisnewreality.Withinthisdebate,itisalsoimportantnottoframecheatingsolelyas“individualbaddecisions”thatrequirepolicing;butalsotorecognizeitasa

collectiveactionproblemthatcomesdowntohowwedesigninstitutionsandassessmentprocessessothatstudentsareencouragedtolearn,and