教育中的生成式人工智能：教育者與專家觀點(diǎn) Generative AI in education Educator and expert views

GenerativeAIineducation

Educatorandexpertviews

January2024

Authors:TheOpenInnovationTeamandDepartmentforEducation

1

Contents

Listoffigures2

Introduction3

Keyfindings4

Methodology6

Qualitativefieldwork6

Quantitativesources7

Detailedfindings8

Responseandadoption8

Applicationandimpact13

OpportunitiesforGenAIineducation19

Challengesandconcerns21

Recommendations29

Annex1:Casestudies32

Annex2:Educatorinterviewees35

2

Listoffigures

Figure1AdoptionofGenAIforeducationamongschoolteachers 9

Figure2AdoptionofGenAItoolsamongonlinechildren 11

Figure3AdoptionofAItools(incl.ChatGPT)forschoolworkamongpupils,asreported

bypupilsandparents 12

Figure4GenAIapplicationsamongstthesubsetofschoolteachersandleadersusingit

14

Figure5ApplicationofAItools(incl.ChatGPT)amongthesubsetofsecondarystudents

(years7to13)whohaveusedit 18

Figure6ReasonsforschoolteachersandleadersnotusingGenAIintheirrole 22

Figure7Extenttowhichschoolleadershavealreadyorplantomakechangesbasedon

GenAItools 24

Figure8Levelsofteacherandleaderconfidenceinadvisingpupilsaboutappropriate

useofAI 25

Figure9Viewsofteachersontheimpactoftechnologyinschoolsonpupilattainment 28

3

Introduction

Overthelastyear,interestinanduseofgenerativeartificialintelligence(GenAI)has

rapidlyincreased.GenAIusesfoundationmodels,includinglargelanguagemodels

(LLMs),trainedonlargevolumesofdata.NotableGenAIfoundationmodelsare

OpenAI’sGPT-3.5andGPT-4,whichunderpinthechatbotsChatGPTandBingChat.

1

Thesetoolscanbeusedtoproduceartificiallygeneratedcontentsuchastext,audio,

code,imagesandvideos.OtherexamplesofGenAItoolsincludeGoogleBard,ClaudeandMidjourney.Thistechnologyisalsoincreasinglybeingintegratedwithinotherdigitaltools.

AlthoughGenAIisnotnew,recentadvancesintheunderlyingtechnologyandgreateraccessibilitymeanthatthepubliccannowuseitmoreeasily.Thisposesopportunitiesandchallengesfortheeducationsector.

TheDigitalStrategyDivisionintheDepartmentforEducation(DfE)askedHM

Government’sOpenInnovationTeam(OIT)toexploretheopportunitiesandrisksforGenAIineducation.

Thisreportcontainsinsightsfrominterviewswithteachersandeducatorsat23

educationalinstitutions,14interviewswithexpertsfromacademiaandtheeducation

technology(EdTech)industry,arangeofquantitativedatasources,andkeythemesfromacademicandgreyliterature.WehavealsodrawnontheDfE’s

CallforEvidence

summaryofresponses

whererelevant.

Thereportcovers:

.HowthesectorhasrespondedtoandadoptedGenAItechnology.

.ApplicationsandopportunitiesforGenAIineducation.

.ReportedimpactandbenefitsofGenAIuseineducation.

.BarrierstoadoptionandrisksthatGenAIpresentsforeducation.

.SupportthesectorwouldliketoreceivefromtheDfEandgovernment.

Whenusedappropriately,technology(includingGenAI),hasthepotentialtoreduce

workloadacrosstheeducationsector,andfreeupteachers’time,allowingthemtofocusondeliveringexcellentteaching.Wewanttocapitaliseontheopportunitiestechnology

1AtthetimeofwritingtherewerethreeversionsofChatGPTavailable:1)ChatGPT-3.5,afreelyavailablechatbot;2)ChatGPTPlus,whichusesGPT-4andwasmadepubliclyavailableinMarch2023viaa￡16-permonthsubscription;and3)ChatGPTEnterprise,aimedatbusinesses,whichlaunchedinAugust2023.

4

likeGenAIpresentsforeducationaswellasaddressingitsrisksandchallenges,

includingitspotentialtotailoreducationalmaterials,andsupportstudentswithSEND.

TheDepartment

publishedapositiononGenAIinEducation

on29March2023.Along-sidethisposition,the

CallforEvidence

onGenAIinEducation,andtheGenAI

Hackathonsproject,weareinvestingupto￡2millioninOakNationalAcademyto

improveandexpandtheirAItoolsforteachers.Wehavealsoprovidedafurther￡137milliontotheEducationEndowmentFoundationtoencourageinnovativeandeffective

evidence-basedteaching,includingusingtechnologysuchascomputeradaptivelearningandAI.

Technologyworksbestasatoolusedbygreatteachers,anditisimportanttotakea

joined-uppedagogicalapproach.Technology,includingGenAI,isnotacatchallsolutiontoeducationalchallengesandcouldneverreplacethevaluablerelationshipbetween

teachersandpupils.Similarly,skillslikehandwritingwillcontinuetobeimportantin

children’sdevelopmentandschoolinginEngland.Buttechnologycansupportand

augmentbrilliantteachers’teaching.Itsuseintheclassroomshouldbeinformedby

evidenceandbestpractice,whichiswhytheDepartmentcontinuestobuildtheevidencebaseforthistechnology.

Schools,collegesanduniversities,aswellasawardingorganisationsneedtoremain

awareoftherisksofnewtechnologiesandcontinuetotakereasonablestepstomitigatethemandpreventmalpractice.Mitigationsforpotentialmalpracticealreadyexistwithin

thesystem,includingin-personexamswherepupilsdon’thaveaccesstotheinternetandthereforecannotuseAI.Inaddition,itisstandardpracticethatexamsarehandwritten.

TheJointCouncilforQualifications

publishedguidance

inMarch2023whichremindsteachersandassessorsofbestpracticeinpreventingandidentifyingpotential

malpracticeinnon-examinedassessment,applyingitinthecontextofAIuse.Ofqualspeakregularlywithexamboardsaboutrisks,includingmalpracticerisks.

Keyfindings

TeachersandexpertsacknowledgethatGenAIcouldhaveatransformativeimpactoneducation.Fromhelpingteacherssavetimebyautomatingtasks,toimprovingteachingeffectivenessbypersonalisinglearningforstudents,thereissignificantpotentialfor

GenAItobenefitthesector.Atthesametime,thereisconsiderableconcernabouttherisksitpresents,aswellasscepticismaboutwhetherthesecanbemitigated.

UseofGenAIamongteachersandstudentshasrapidlyincreasedoverthelastyear.ByNovember2023,42%ofprimaryandsecondaryteachershadusedGenAIintheirrole

(anincreasefrom17%inApril).PupilsandstudentsmaybeusingGenAImorethantheirteachers.74%percentofonline16-24yearoldsintheUKhaveusedaGenAItool.

5

EvidenceonGenAIuseinaneducationalcontextshowsfiguresrangingfrom14%to67%ofstudentshavingusedGenAIforschoolworkandstudies.

EducatorsarealreadyrealisingthebenefitsofusingGenAI,namelyinhelpingthemsavetimebycreatinglessonresources,planlessonsandstreamlineadministrativeprocesses.Reportedbenefitsforstudentsincludeincreasedengagementandbettersupportfor

studentswithspecialeducationalneedsanddisabilities(SEND)bypersonalisinglearningmaterials.

ThereiswidespreadrecognitionoftherisksGenAIpresentsforeducation.Thereis

considerableconcernaroundGenAI-enabledacademicmalpractice,studentover-

relianceonGenAI,aswellasethical,safetyanddataprivacyrisksofuse.ThepotentialforGenAItowideneducationalinequalitieswasraised.ThereisalsoconcernthatthebenefitsofGenAIforeducationwillneverbefullyrealisedduetobarrierstoadoption.

LackofknowledgeabouthowtouseGenAI,andpoordigitalskillsandinfrastructurearelimitingfurtheruseamongteachers.Negativemediacoverageaffectssometeachers’perceptionsofGenAI,aswellasraisesconcernaboutthethreatofAItoteacherjob

security.

ThereisappetiteforgovernmentsupporttoensureGenAIadoptionineducationissafe,effective,andalignswithgoodpedagogy.ThereislittlerobustevidenceontheimpactofGenAIineducation,withexpertsemphasisingtheneedtobuildanimpartialevidence

basetobetterunderstandtheimpactofGenAItoolsoneducation.Suggestionsfortheroleofgovernmentincludedsharingguidanceandbestpracticeandensuringstudent

dataprotectionandprivacy.AsthesectoradaptstoGenAI,expertsandeducatorsalsohighlighttheneedforalonger-termstrategyforartificialintelligence(AI)ineducationthatisfuture-proofedtokeeppacewiththeevolvingnatureofthistechnology.

6

Methodology

Thisreportdrawsonprimaryqualitativeandquantitativeresearchaswellaspublishedinformationfromarangeofgovernment,academicandprivatesectororganisations.

Themethodologiesunderpinningthefindingsarevaried,includingweighted

representativeonlinepanels,non-representativesamples,andannualsnapshotsurveys.

Whilesourceshavebeenselectedtoprovidethemostup-to-dateassessmentavailable,thereareclearlimitationsinwhatcanbeconcluded.GenAIisanemergingtechnology

whichisrapidlyevolvinganditsuseineducationischanging.Thismeansthatsomefiguresandfindingspresentedinthisreportmaybecomeoutofdatequickly.

Itisalsoimportanttonotethatsomeofthefindings,particularlythosedrawnfrom

qualitativeresearch,areonlyrepresentativeofthosewhoparticipatedandcannotbetakentorepresenttheviewsofallindividuals,professionalsandorganisations.

Qualitativefieldwork

InterviewswithexpertsandteacherswereconductedfromMaytoSeptember2023.TheOITheardfrom14academicexpertsandEdTechdevelopersoverMayandJune2023.

.

.

.

.

.

.

.

.

.

.

.

.

.

Prof.RoseLuckin,

UniversityCollegeLondon

Prof.SteveWatson,

UniversityofCambridge

Prof.MikeSharples,

TheOpenUniversity

Prof.RebeccaEynon

,UniversityofOxford

Dr.JunLiu

,UniversityofUlster

Prof.DonPassey

,UniversityofLancaster

Dr.AndrewRogoyski

,UniversityofSurrey

Prof.SteveHiggins,

DurhamUniversity

Prof.TimFawns,

MonashUniversity

MichaelWebb

and

SueAttewell,

Jisc

Dr.AlinavonDavier

,UniversityofOxfordandDuolingo(EdTechindustry)

Dr.RajeshwariIyer,

sAIaptic(EdTechindustry)

YvonneSoh,

NoodleFactory(EdTechindustry)

7

Educatorsworkingin23educationalsettingsacrossEnglandwereinterviewedover

AugustandSeptember2023.SomeparticipantswererecruitedfromtheDfE’sCallforEvidence.Thisincluded:

.Twoeducatorsfromearlyyears.

.Nineeducatorsfromprimaryschools.

.Eighteducatorsfromsecondaryschools.

.Foureducatorsfromfurthereducation(FE)institutions.

.Threeeducatorsfromhighereducation(HE)institutions.

.TwoeducatorsfromspecialschoolsandSENDprovision.

Pleasenotethatsomeeducatorsrepresentedmultiplestages(seeAnnexBforthefulllistofeducatorsinterviewed).

Quantitativesources

Keysourcesofdatainthisreportinclude:

.DfE–

SchoolandCollegesPanel–April2023wave

.DfE–

Parent,PupilandLearnerPanel–April/May2023wave

.DfE–

ImpactofAIonUKjobsandtraining–November2023

.Ofcom–

OnlineNation2023report–November2023

.OrielSquare–

EducationIntelligence–AIineducationedition

.Deloitte–

DigitalConsumerTrends–August2023

Additionalexternalquantitativesourceswerealsoreviewed.

8

Detailedfindings

Responseandadoption

TheeducationsectorisacknowledgingtheneedtoadapttoGenAItechnology.

AdvancesinAIarelikelytohaveaprofoundandwidespreadeffectontheUKeconomyandsociety.

2

EducationisoneofthetopsectorsexpectedtobeimpactedbyAIand

GenAI,alongsideindustrieslikebanking,researchshows.

3

4

Inadditiontoaugmentingeducatorjobsandtasks,GenAIcouldalsofundamentallyalterhowandwhatpeoplelearnbychanginghowinformationissynthesisedandpresented.

Thesectorisrespondingtowidespreadaccesstothistechnology.Initiallysome

educationalinstitutionsprohibiteduseduetoplagiarismconcerns.However,

acknowledgmentthatthistechnologyisheretostayisdrivingamoreadaptivestance.

GuidancehasbeenprovidedbytheJointCouncilforQualifications(JCQ)forprotectingtheintegrityofqualifications,andRussellGroupuniversitieshaveagreedprinciplesforAIuse.

5

6

UNESCOhaspublishedglobalguidanceonGenAIuseineducation,including

proposingframeworksforregulation.

7

ExpertsandeducatorsacknowledgeGenAI’spotentialtobenefiteducation,suchasbyactingasatooltosupportteacherswiththeirworkloadandenableself-directedand

personalisedlearningforstudents.ThereisalsoastrongsenseGenAIcouldprofoundlydisruptthesector,includingbychangingwhatstudentsneedtolearntopreparethemtoenteranAI-enabledworkforce.

8

However,thereisuncertainty,concern,andscepticism.Thereisahistoricalpatternof

hypethatEdTechwillimprovetraditionaleducation.

9

Yetdecadesofresearchshowsthattheuseoftechnologyinschoolingresultsinmixedandunevenoutcomes,andthiswasbroughtsharplyintofocusduringtheCOVID-19pandemic.

10

ArecentDfEsurveyshowsthatteachersaredividedintheirviewsaboutwhethertechnologyusedinschoolshasa

2DepartmentforEducation(2023).

“TheimpactofAIonUKjobsandtraining”

.

3

/featured

-insights/mckinsey-explainers/whats-the-future-of-generative-ai-an-early-view-in-15-charts

4DepartmentforEducation(2023).

“TheimpactofAIonUKjobsandtraining”

.

5JointCouncilforQualifications(2023).

“AIUseinAssessments:ProtectingtheIntegrityofQualifications”

.6RussellGroup(2023).

“RussellGroupprinciplesontheuseofgenerativeAItoolsineducation”

.

7UNESCO(2023).

“GuidanceforgenerativeAIineducationandresearch”

.

8

/commentisfree/2023/jul/14/ai

-artificial-intelligence-disrupt-education-creativity-critical-thinking

9UNESCO(2021).

“Digitaltechnologyandthefuturesofeducation–towards‘non-stupid’optimism”

.

10UNESCO(2023).

“Aned-techtragedy?Educationaltechnologyandschoolclosuresinthetimeof

COVID-19”

.

9

positiveornegativecontributiontopupilattainment.

11

GenAIalsointroducesmanynewrisksthatneedtobemanaged.

12

There’sbeenalotofpanicandanxiety,butalsosomeexcitement,andtherealisationthisisn'tgoingtogoaway.Itwillfundamentallychangeeducation.–RoseLuckin,UniversityCollegeLondon

AdoptionofGenAIamongteachershasrapidlyincreased,withtwoinfiveteachersnowhavingusedGenAIintheirrole.

AccordingtoasurveyfromTeacherTappinNovember,42%ofprimaryandsecondaryteachershavenowusedGenAItohelpthemwithschoolwork(seeFigure1).

13

Thishasincreasedfrom35%inAugustand17%inApril.

14

15

Incomparisontothewiderpublic,aroundaquarteroftheUKconsumers(aged16-75)hadusedaGenAItoolinJune

2023.

16

Figure1AdoptionofGenAIforeducationamongschoolteachers

NovemberAugust

Yes,outsideofwork

Yes,tohelpwithschoolworkYes,inalesson

IhavenotusedAI

Idon'tknowwhatAIis

29%

28%

42%35%

7%

6%

44%

50%

1%

2%

Source:TeacherTapp.Primaryandsecondaryteachers.QuestionaskedinAugust(n=9,138)andinNovember(n=9,275).Resultsweightedtoreflectnationalteacherandschooldemographics.Respondentscouldselectmultiple“yes”options.

11DepartmentforEducation(2023).

“Technologyinschoolssurvey2022to2023”

.

12GovernmentOfficeforScience(2023).

“FutureRisksofFrontierAI”

.

13

https://teachertapp.co.uk/articles/how

-to-improve-behaviour-wellbeing-and-how-youre-using-ai-in-schools/

14

https://schoolsweek.co.uk/chatgpt

-one-in-three-teachers-use-ai-to-help-with-school-work/15OrielSquare(2023).

“EducationIntelligencereport”

.

16Deloitte(2023).

“DigitalConsumerTrends2023”

.

10

Educatorsinterviewedreportedusingarangeofnon-specialisedGenAItools,includingChatGPT,GoogleBard,Bing,DALL-E,Midjourney,Canva,andMicrosoftDesigner.

SomewerealsousingeducationspecificGenAItools,suchasTeachMateAI,AskArborandMemrise.

Therewassignificantvariationinthefrequencyofuseamongtheeducatorsinterviewed.SomereportedusingGenAItoolsdaily,andthatusewaswidespreadamongcolleagues.OthershadonlyusedGenAIafewtimesandwerenotawareofcolleaguesusingit.

Educatorswhoaremen,youngerandthoseinsecondaryschoolsaremorelikelytobeusingGenAI.

DfE’ssurveyofschoolandcollegeteachers(conductedinApril2023)showedthatthosewhoweremorelikelytohaveusedGenAIwere:

.Workinginsecondaryschools:14%ofsecondaryschoolteacherscomparedto9%inprimaryschools.23%ofcollegeeducatorshadusedGenAI.

.Male:18%ofteacherswhoaremencomparedto9%ofteacherswhoarewomen.

.Younger:15%ofteachersaged18-34hadusedGenAIcomparedto9%ofthoseaged45-54.

17

IndependentschoolteachersweremorelikelytouseGenAIintheirrolescomparedto

thoseinstateschoolsandtoreportthattheyhadreceivedinformationaboutusingAI

technologiesinschool,otherresearchshows.

18

17%ofteachersinindependentschoolshadusedGenAItohelpwiththeirworkcomparedto9%ofteachersatstateschools.

ThisalignswithtrendsinawarenessofGenAIamongtheUKpublic:

.Men:60%ofmenhadheardofGenAIcomparedto46%ofwomen.

.Youngpeople:73%ofunder35shadheardofGenAIcomparedto27%of65-75s.

19

PupilsandstudentsmaybeusingGenAImorethaneducators.

Ofcomreportsthat74%percentofonline16–24-year-oldsintheUKhaveusedaGenAItool.

20

Amongyoungerchildrenuseissimilarlyhigh,with79%ofonlineteenagers(aged13-17)and40%ofonline7–12-year-oldsreportingthattheyhadusedChatGPT,

SnapchatMyAI,MidjourneyorDALL-E(seeFigure2).ThemostcommonlyusedGenAItoolamongchildrenisSnapchatMyAI.

17

18

19

20

DepartmentforEducation(2023).

“SchoolandCollegePanel–April2023.ResearchReport”

.OrielSquare(2023).

“EducationIntelligencereport”

.

Deloitte(2023).

“DigitalConsumerTrends2023”

.

Ofcom(2023).

“OnlineNation2023Report”

.

11

Figure2AdoptionofGenAItoolsamongonlinechildren

Age13-17Age7-12

Used

any

oftheseAItoolsSnapchatMyAI ChatGPT DALL-EMidjourney

79%

40%

72%

30%

29%

19%

7%

6%

5%

7%

Source:CHILDWISEsummeromnibus2023.Servicesusedinthepastsixmonths:FieldworkconductedJune-July2023.AsreportedinOfcomOnlineNation–2023Report.

WhenitcomestousingGenAIforschoolworkspecifically,educatorsinterviewed

reportedstudentscommonlyusingChatGPT,Bard,Midjourney,andCanvatosupporttheirstudiesorinlessons.OnereportedastudentusingSnapchatMyAItoask

questionsabouttheirhomework.

DfE’ssurveyofsecondaryschoolpupilsshowed14%usingAItoolsforschoolwork(seeFigure3).However,othersurveysshowfiguresof67%ofsecondaryschoolpupilsusingGenAItoolsforschoolworkandaroundhalfofHEstudentsusingChatGPTtosupporttheirstudies.

21

22

UnderstandingthetruefigureofGenAIuseamongstudentsforschoolworkandstudies,includingusethatmayconstituteacademicmalpractice,ischallengingtodeterminedue

tousebeinghardtodetectandnotalwaysreportedbythoseusingit.However,

instancesofacademicmalpracticeduetoAIwereanecdotallyreportedbytheexpertsandeducatorsinterviewed,aswellasinDfE’sCallforEvidenceSummaryof

Responses.

23

StudentsaremorefamiliarwithGenAIthanstaff.StaffknowthatChatGPTexists,butalmostallofouryear10shaveexploredhow

21

/news/2023/artificial

-intelligence-in-education

22

/company/news/survey

-finds-uk-students-using-ai-to-support-studies/

23DepartmentforEducation(2023).GenerativeAIineducation–

“CallforEvidence:summaryof

”

responses.

12

ChatGPTandotherGenAIcanbeusedethically.–PeteDring,FulfordSchool

Parentslackclarityontheirchildren’suseofAItools.DfE’ssurveyofparentsof

secondaryschoolpupilsshowedthat37%wereunsurewhethertheirchildusedAIforschoolwork(seeFigure3).

Figure3AdoptionofAItools(incl.ChatGPT)forschoolworkamongpupils,asreportedbypupilsandparents

Pupils(years7-

13)

Parents

(secondary)

73%

37%

14%

12%

10%

54%

YesNoDon'tknow

Source:DfEParent,PupilandLearnerPanel(April/May2023).Pupilsinyears7to13(n=3,238),secondaryparents(n=1,738).

SomeinstitutionsareactivelymanaginghoweducatorsandpupilsuseandexperimentwithGenAI.

Someeducatorsreportedthattheirinstitutionhadbannedteacherand/orpupiluseof

GenAI.Thiswasprimarilyenforcedthroughserverrestrictionsorpolicies.However,

educatorsandexpertsinterviewedgenerallyviewedoutrightbansasshortsighted,duetothesensethatthistechnologyiswidelyaccessibleandwillbeincreasinglyintegratedintoexistingtoolsandplatformsusedineducation(e.g.MicrosoftCopilot).

Otherinstitutionsweremanagingadoptionattheirinstitutionsbyconductingtoolrisk

assessments,allowingasmallgroupofteacherstoexperimentwith“approved”GenAI

tools,orpilotingtools.OthereducatorsreportedtheirinstitutionshadminimaloversightofhowtheywereusingGenAItools,andfelttheyhadautonomytoexperiment.

Oneeducationalinstitutioninterviewed,BoltonCollege,isexperimentingwithintegratingGenAItechnologyintothecollege’sexistingdigitaltools.Thecollegehasintegrated

GenAIintoastudentsupportchatbottoenhanceitsperformance,andisapplyingGenAItoitsinternalassessmentplatform,FirstPass.

24

24

https://www.fenews.co.uk/fe

-voices/whats-next-for-bolton-colleges-ada-service/

13

EarlyadopterinstitutionsrelyonGenAIadvocatestodriveuse,andtendtobemoredigitallymature.

EducatorswhowereusingGenAIreportedanumberoffactorsthattheyconsideredtosupporttheirGenAIuseandencouragewideradoptionattheirinstitution.

Champions:EducatorsusingGenAIreportedhavingapersonalinterestinitandtrainedthemselvesonhowtouseit,includingintheirsparetime.Socialmediaplatforms,

includingLinkedInandInstagram,werecitedasusefulsourcesofinformationandtips.

TheseGenAIeducator“champions”advocateforGenAIuseandsupportwiderknowledgeandskillssharingattheirinstitutions.

Digitalmaturityoftheinstitution:InstitutionswithhighlevelsofGenAIuseoftenhadstrongITinfrastructure,resourcededicatedtoITtrainingandsupport,andwidespreaduseofEdTech.InstitutionsalsohadprocessesinplacetoreviewandapproveGenAItools(e.g.conductriskassessments)inlinewithotherEdTechtoolsanddigital

infrastructure.Someinstitutions,suchasmulti-academytrusts(MATs),managedtoolapprovalcentrally.

Collaborationwitheducators:Earlyadoptereducatorsreportedcollaboratingwithcolleaguestoshareknowledge,includingrunningtrainingsessionsforeachotherorinformallydiscussingideasforGenAIuse.KnowledgesharingacrosssomeMAT

networkshelpedfacilitateadoption.

Seniorsupport:EducatorsusingGenAIoftendescribedtheirinstitutionalcultureasbeinginnovativeandsupportiveofuseofnewtechnologies.SomereportedseniorleadershipwhoactivelyendorsedGenAIorgaveothersautonomytoexperiment.

Applicationandimpact

ApplicationofGenAIamongteachers

ThemostcommonapplicationofGenAIamongschoolandcollegeteachersiscreatingeducationalresources.

EducatorsarecommonlyusingGenAIinrelativelylow-riskways,suchascreating

contentforlessonsandforadministrativesupportive(e.g.writinglettersandemailstoparents,developingmarketingresources)(seeFigure4).Fewereducatorswere

experimentingwithusingGenAIformarkingandassessment.OveroneintenwereusingGenAItosupportlearnerswithSEND.

14

Figure4GenAIapplicationsamongstthesubsetofschoolteachersandleadersusingit

Creatinglessons/curriculumresources

Planninglessons/curriculumcontent

Communicatingandengagingwithparents/carers

Deliveringlessons(e.g.usingAItools

duringliveteachinginclass)

Formativeassessments(e.g.giving

feedbackormarking)

Draftingpolicydocuments

SupportingpupilswithSEND

Collaboratingandsharingresourceswithotherteachers

SummativeassessmentInteachertraining

Other

Don'tknow

42%

17%

16%

13%

13%

13%

12%

11%

5%

5%

1%

62%

Source:DfESchoolandCollegesPanel(April2023).LeadersandteachersusingGenAIintheirrole(n=280).

Otherapplicationsreportedbyeducatorsincluded:

.Researchandwritingaid:UsingGenAItoolstosupportresearchtasks,includingresearchingatopicorconceptandsummarisingarticles,booksandvideos.GenAItoolswerealsousedtotranscribeortranslatecontent,aswellasproofreadand

editwrittencontent.

.GenAIskillsandAIliteracy:EducatorsusedGenAItoolstodelivertrainingtootherstaffandpupilsonwhatGenAIisandhowtouseGenAItools,aswellasraisingawarenessoftherisksandlimitationsof