《英国文化教育协会:2023人工智能与英语教学:为未来做好准备研究报告(英文版)(76页).pdf》由会员分享,可在线阅读,更多相关《英国文化教育协会:2023人工智能与英语教学:为未来做好准备研究报告(英文版)(76页).pdf(76页珍藏版)》请在三个皮匠报告上搜索。
1、www.britishcouncil.orgArtificial intelligence and English language teaching:Preparing for the futureDr Adam Edmett Neenaz IchaporiaDr Helen CromptonRoss Crichton English ProgrammesISBN:978-1-915280-35-0 DOI:https:/doi.org/10.57884/78ea-3c69 British Council 2023 1 Redman Place,London E20 1JQ,United K
2、ingdom www.britishcouncil.orgCite this report:Edmett,A.,Ichaporia,N.,Crompton,H.,&Crichton,R.(2023).Artificial intelligence and English language teaching:Preparing for the future.British Council.https:/doi.org/10.57884/78EA-3C69Photo credits:All photos copyright British Council unless otherwise mark
3、ed.Front cover image Mat WrightArtificial intelligence and English language teaching:Preparing for the futureAbbreviationsAI artificial intelligenceCALL computer-aided language learningCEO chief executive officerCPD continuing professional development ELT English language teachingELT/L English langu
4、age teaching and learningEU European Union GenAI generative AI GPT Generative Pre-trained TransformerL2 second languageLGC learner-generated contextLLM large language modelSDG Sustainable Development GoalsSTEM science,technology,engineering and mathematics TOEFL Test of English as a Foreign Language
5、TOEIC Test of English for International CommunicationUNESCO United Nations Educational,Scientific and Cultural OrganizationAcknowledgementsThis publication would not have been possible without the contributions of our 19 interviewees,who come from various walks of life and countries.By consenting to
6、 be interviewed,they volunteered their valuable time to our project,and their views have been invaluable in shaping Part III of this publication.A collective thanks to the 1,348 anonymous English language teachers from around the world who took time to respond to our survey for this publication.Thei
7、r voices helped us capture the views of teachers worldwide,arguably the most important stakeholders for this report(see Part II).Thanks also to the lead researcher on the systematic review that forms the basis of Part I of this publication,Dr Helen Crompton,as well as Senior Research Associate at OD
8、UGlobal Diane Burke.We would also like to thank our reviewer Amy Lightfoot,Director Insight and Innovation,British Council who also contributed the foreword to this piece.Finally,thanks to the British Councils English Programmes EdTech Innovation and English Connects teams.Their input to discussions
9、 around the topic and help with the teacher survey were invaluable.About the British CouncilWe support peace and prosperity by building connections,understanding and trust between people in the UK and countries worldwide.We work directly with individuals to help them gain the skills,confidence and c
10、onnections to transform their lives and shape a better world in partnership with the UK.We support them to build networks and explore creative ideas,to learn English,to get a high-quality education and to gain internationally recognised qualifications.We work with governments and our partners in the
11、 education,English language and cultural sectors,in the UK and globally.Working together we make a bigger difference,creating benefit for millions of people all over the world.We work with people in over 200 countries and territories and are on the ground in more than 100 countries.In 202223,we reac
12、hed more than 80 million people directly and 791 million people overall,including online,through broadcasts and publications.Founded in 1934,we are a UK charity governed by Royal Charter and a UK public body.Image Mat WrightForeword 09Introduction 10Part I:Background:What the literature says 13i.Dem
13、ographic trends 14ii.Educational benefits of AI in ELT 15iii.Challenges of AI in ELT 18iv.Implications for practice 19v.Areas for future research 20Part II:The survey:What teachers say 21i.About the survey 22ii.Contributing teachers 22iii.How teachers are using AI in ELT 24iv.Teachers views on AI in
14、 ELT 26v.AI in ELT:the present 26vi.AI in ELT:the future 32vii.Summary 36Part III:The interviews:What our key witnesses say 37i.Approach 38ii.Emerging themes 39 a.Definitions 39 b.Pedagogy 40 c.Big Tech and neoliberalism 42 d.Replacing humans 44 e.Relevance for ELT 47 f.Bias 49 g.Teacher readiness 5
15、2 h.Motivation 54 i.Inclusion 55 j.Assessment 57 k.Ethics frameworks and regulation 59iii.Next steps and final thoughts 60iv.Interviewee profiles 61About the authors 67References 68Appendices 71ContentsArtificial intelligence and English language teaching:Preparing for the future 07Image InPress Pho
16、tographyArtificial intelligence has captured the worlds imagination,generating countless headlines and causing heated debates.These discussions are very live among those working in education what impact will AI have on how our learners gain knowledge and develop skills?What impact will it have on ho
17、w we recruit and train our teachers?Will teachers ultimately be replaced by technology?Language teaching in particular presents multiple opportunities for the integration of AI-powered technologies.Even prior to the development of generative AI tools like ChatGPT,we have seen many successful applica
18、tions developed using AI to create adaptive learning pathways for language learners.Generative AI tools now provide incredible potential for language practice.However,realising that potential requires motivation and skills from learners,teachers and many other stakeholders.It is clear that there are
19、 both barriers and risks which need to be explored,and the voices of those whom these technologies are aimed at must be listened to and carefully considered.This publication aims to do just that.It recognises the changes that AI is bringing to the world of English language teaching and seeks to unde
20、rstand the implications,and peoples feelings about them,through the examination of three sources of evidence:the literature,teachers and a group of expert key witnesses who reflect on the possibilities and considerations unfolding before us.At the British Council,we contribute to the development of
21、sustainable education systems.Recognising the opportunities it brings,we have a particular focus on supporting the teaching,learning and assessment of the English language.All of this work is underpinned by evidence,and by listening carefully to our partners and the educators we work with.By doing t
22、his,we help to improve the quality of classroom practice around the world,with intentional use of education technologies to provide support where this is appropriate.We are excited to learn about how AI might help to move our field forward,and mindful of the need to prepare and support those who wil
23、l use it to mitigate risks and remove obstacles where we can.This publication gives us all important insight into current attitudes of the relationship between artificial intelligence and English language teaching,and how teachers are using the new technologies.It also raises key questions for furth
24、er consideration and points to critical next steps that will help to promote its safe integration into our classroom practice.Amy Lightfoot Director Insight and Innovation English Programmes British Council ForewordArtificial intelligence and English language teaching:Preparing for the future 09The
25、discussion around how artificial intelligence(AI)will impact every field of human activity has gained traction following developments in conversational AI.Since late 2022,AI-powered technology,such as ChatGPT and its iteration GPT-4 and similar large language models(LLMs),has taken the world by stor
26、m.These have been accompanied by the release of other generative AI tools that can rapidly produce language,images and computer code with remarkable ease.There is increasing recognition that these emerging technologies present immediate as well as far-reaching opportunities,challenges and risks to e
27、ducation systems(UNESCO,2023).However,in the context of implementing artificial intelligence in the educational sector,the existing research and guidelines are relatively limited,and there is a need for more extensive exploration of these issues.In the specific field of English language teaching and
28、 learning(ELT/L),although there are now a wide range of resources available for teachers on classroom-based use of AI tools(such as blogs,webinars,how-to guides),there is a need for a deeper engagement with the opportunities,issues and challenges AI presents.This publication was therefore commission
29、ed by the British Council to contribute to emerging research specifically around the impact of AI on the teaching of English in education systems.As a first step towards understanding the current scenario,education technology experts at the British Council and Dr Helen Crompton,Director of the Resea
30、rch Institute of Digital Innovation in Learning(RIDIL)ODUGlobal,sought to investigate the current state of research evidence around the use of AI in ELT/L.The team initially completed a first-of-its-kind systematic literature review that maps out the last ten years of research focused on the topic.P
31、art I of this publication summarises the main findings from this review,including the main affordances and challenges that emerged and the implications for practitioners.For readers who would like more information about how we approached the IntroductionIn the specific field of English language teac
32、hing and learning.there is a need for a deeper engagement with the opportunities,issues and challenges AI presents.Artificial intelligence and English language teaching:Preparing for the future 10Image Mat Wrightsystematic review,the coding we employed and the detailed findings,please access this he
33、re.1Having studied the literature to establish how AI is currently being used for English language teaching(ELT)and taking into account the fact that the majority of this peer-reviewed literature was written prior to the widespread availability of more recent generative AI tools it was felt that any
34、 commentary on the current situation and the possible future of AI in ELT required capturing a wider range of stakeholder voices.Therefore,this report builds on our learning from the systematic review and incorporates views and opinions from across the world.Following the brief overview of the liter
35、ature in Part I,in Part II we present the results of a global survey of 1,348 English language teachers from 118 countries,capturing their views through both numerical analysis of their aggregate responses as well as the sentiments of individuals captured through more in-depth commentary that many o
36、f them provided.Next,in Part III,we explore the major themes emerging from 19 in-depth interviews conducted with our key witnesses:practitioners and decision makers from a range of geographies,including teachers,government representatives,researchers from higher education,representatives of private
37、language schools,and ELT and EdTech sector experts.Through this publication,we aim to have a multiplicity of voices represented in the conversation on AI in ELT and its future in our field.It is critical that we explain how we understood the term AI for this work,as it is used in a variety of differ
38、ent ways by different people.AI can be defined as computer systems that simulate human intelligence(Sindermann et al.,2021)and can learn,understand and remember human language(Xiaohong&Yanzheng,2021).The literature we analysed encompassed various AI technologies and systems:1)learner-facing,used by
39、pupils to learn,2)teacher-facing,used by teachers to help in teaching activities,for example grading,and 3)system-facing,which is used by administrative staff to manage and examine pupil data(Pokrivkov,2019).Crucially,use of the term AI remains intensely problematic as it is used to refer to a range
40、 of different systems.Where possible,in this publication,we refer to more specific forms of AI such as generative AI(GenAI)or LLMs.In Part III,the need for more nuanced definitions of AI is explored in more depth.1 https:/www.britishcouncil.org/sites/default/files/ai_in_english_language_teaching_sys
41、tematic_review.pdfArtificial intelligence and English language teaching:Preparing for the future 11Image InPress PhotographyPart IBackground:What the literature says Background:What the literature saysA scan of the existing literature indicated that an up-to-date,comprehensive study was needed to pr
42、esent the current use of AI for ELT/L across learners of all ages.The resulting systematic review,commissioned by the British Council,was one of the first formal studies to examine how AI has been used across all ages of learners who are studying English as a second or additional language.It involve
43、d gathering research studies on AI in ELT that fulfilled specific criteria and carefully analysing these for trends and patterns.At the end of the search selection process,43 peer-reviewed articles met the inclusion criteria and were studied in detail.2 The overarching question for this review of th
44、e literature was How is artificial intelligence being used for English language teaching and learning?This section summarises our key findings.Demographic trends Asia is currently at the forefront of AI in ELT research,with over two-thirds(72 per cent)of the articles coming from Asia,19 per cent fro
45、m China.More studies are now being conducted on AI in ELT than there were five or ten years ago,mirroring the recent rise in the number of AI tools available and public interest in AI.Interestingly,the findings show a significant gap in peer-reviewed research on AI in adult ELT outside of the formal
46、 education system.The majority of studies focus on learners in higher education.This perhaps echoes the general demographic shift towards people learning English at younger ages(Trajectory Partnership,2018).Image Mat Wright2 For more detailed information about the methodology,please see here:https:/
47、www.britishcouncil.org/sites/default/files/ai_in_english_language_teaching_systematic_review.pdfArtificial intelligence and English language teaching:Preparing for the future 14Educational benefits of AI in ELTThe review identified five key areas in which AI is being used in ELT:for the development
48、of speaking,of writing and of reading skills,to support pedagogy and for self-regulation.Interestingly,among the language skills,listening did not emerge as an area where AI is being used for support.Pronunciation was the key sub-skill revealed in the studies related to the use of AI in speaking,wit
49、h a variety of AI-powered systems and programs available for learners.For instance,a study with Taiwanese leaners by Liu and Hung in 2016 found that the use of AI and the visual representation of the pitch as a spectrogram significantly improved learners pronunciation by reducing the flatness of pit
50、ch and intonation patterns.Pedagogy or teaching methods in relation to teaching speaking also emerged as an area of interest.AI was used as a conversational partner,a language coach and in a multimodal capacity.For example,Dizon and Tang(2020)had learners converse with Alexa,a personal voice assista
51、nt.They found that it promoted meaningful interactions,supported vocabulary acquisition,improved language skills and provided interesting,enjoyable learning.Other studies highlighted the use of coaching and multimodal systems(employing multiple ways to present information,such as text,images,audio a
52、nd video).For instance,in a study by Shivakumar et al.(2019)in a higher education setting,learners were provided with an AI coach that tailored instruction to each learners learning patterns and needs,resulting in the ability to speak more fluently using consistently accurate language structures.Oth
53、er technologies used for improving speaking skills included using AI for speech recognition,adaptive learning,automatic speech analysis and voice assistance.One example is where Kazu and Kuvvetli(2023)developed an AI-supported pronunciation model for Turkish learners.This system helped learners prac
54、tise,record and react to learners pronouncing words,resulting in longer retention of the vocabulary and significant benefits in learning consonant and vowel sounds.SpeakingImage Mat WrightAI use in writing related mainly to vocabulary learning and grammar.For example,Lo(2023)found that access to neu
55、ral machine translation programs resulted in learners vocabulary improvement,especially when specialised or unambiguous expressions were involved.Another common use of AI in writing is the use of AI grammar checkers.For instance,a study by Dizon and Gayed(2021)in higher education found that learners
56、 using the AI-powered tool Grammarly made fewer grammatical errors and wrote with more lexical variation than learners who did not.Only one pedagogical focus,to support giving feedback,emerged in AI use for writing skills.Studies looking at pedagogy in writing were often connected to AI tools provid
57、ing feedback via spelling and grammar checkers,along the lines of Dizon and Gayeds(2021)study with Grammarly(above).Nazari et al.(2021)also examined the use of Grammarly as a feedback tool for English language learners.They reported positive outcomes,with an improvement in behavioural,emotional and
58、cognitive engagement,as well as self-efficacy in writing.A variety of AI technology tools were used to support writing skills,including grammar checkers,writing assistants,translation tools and pattern checkers.A study by Chon et al.(2021)with South Korean college learners explored the use of machin
59、e translation as a reference tool for second-language(L2)writing,finding that using Google Translate helped less-skilled learners to display a level of writing proficiency that was not significantly different from that of skilled learners.It also found that machine translation aided learners to prod
60、uce essays with a greater number of lower-frequency,more complex words and higher-quality syntax.WritingAlthough some studies did involve the use of AI for developing the receptive skill of reading,these were far less common than for the productive skills of speaking and writing.Vocabulary was the o
61、nly aspect of developing reading skills that appeared to be a key focus,while only gaming emerged as a specific use to support pedagogy.For instance,Zheng et al.(2015)explored how vocabulary learning in reading occurs during gaming quest-play mediated in English,in the game World of Warcraft(WoW).Th
62、e findings suggest that learners have opportunities to learn vocabulary and understand meaning via games beyond what a textbook or classroom can provide,by contextualising often decontextualised vocabulary.WoW uses AI to provide that context through the inclusion of AI characters(i.e.those not opera
63、ted by a human)and pathfinding navigation algorithms that make the environment dynamic and engaging.ReadingArtificial intelligence and English language teaching:Preparing for the future 16This refers to the methods,strategies and techniques used to facilitate ELT.It is noteworthy that even with the
64、rapid changes in available technology,many conventional forms of pedagogy,such as lectures and explanations,are still in use.Some studies examined multiple approaches that appear to provide a more personalised learning approach.For example,Kim(2022)explored the effects of the pedagogical approaches
65、of score predictions,lectures,explanations and practice tests on Korean learners studying for their Test of English for International Communication(TOEIC).Learners began with a diagnostic assessment in which the AI then used the data to provide lectures,explanations and practice tests at the level r
66、equired by the learner.Lee et al.(2023)explored a learner-generated-context-based(LGC)approach.They defined LGC as the creation and use of digital technology that enables learners to build a learner-generated context and learn within it.This context is derived from data that is collected as the lear
67、ners perform actions and make choices.The system then adapts to the learners and provides them with more content that suits their preferences.The researchers reported that the LGC AI-powered pedagogical approach fostered learners self-autonomous learning experiences.PedagogyLearners emotions,or affe
68、ct,can influence their choices and actions they take.Several studies in our review explored using AI to promote self-regulation the ability to manage and control ones thoughts,emotions,behaviours and physiological responses to achieve personal goals and maintain well-being.This research investigated
69、 the results of engaging learners in active thinking,especially regarding their goals and learning autonomy.What emerged is the trend for AI to allow learners to actively participate in goal setting and become independent learners.For example,Hew et al.(2023)used chatbots in ELT to support learner g
70、oal setting and social presence in fully online activities.This helped learners to clarify their learning goals,create techniques for setting goals and raise awareness of learning strategies in goal setting.In another study,Chen,Hsu et al.(2022)examined robot-assisted language learning,where AI and
71、virtual reality were combined to create a system to use robots as a tool for training English language tour guides to develop a sense of autonomy.The findings of the study identified benefits including increased autonomy,motivation and engagement.Finally,some of the studies we reviewed explored usin
72、g AI to reduce anxiety,which relates to learners feeling anxious about learning English,for example around speaking in public,making mistakes with vocabulary and interacting with others.Chen,Koong et al.(2022)reported that anxiety was reduced when an AI automatic speech recognition tool was used wit
73、h fifth-grade Taiwanese learners.In addition,both akmak(2022)and Chen,Koong et al.(2022)found that their use of AI both raised the learners skills and lowered anxiety.Self-regulationArtificial intelligence and English language teaching:Preparing for the future 17Challenges of AI in ELTRemarkably,the
74、 challenges and risks of AI systems in ELT were not as well reported as its benefits in the research literature we reviewed.However,where they were,there were four main challenges that emerged.Technology breakdowns included technical malfunctions and poor connectivity.One specific technology breakdo
75、wn was incorrect answers given by the AI.Limited capabilities where users required more advanced functionality.For instance,some learners wanted better chatbot capacity and others wanted more natural interactions(Thompson et al.,2018).These limited capabilities led to learners becoming uninterested
76、in using the chatbot.Fear took several forms,including 1)a lack of clarity on how personal information would be stored and shared,2)fear of the unknown,i.e.uncertainty about how the AI was operating,and 3)fear of losing a natural learning environment and,along with it,real emotions connected to lear
77、ning(for example Viktorivna et al.,2022).Standardising languages and ideologies emerged as one of the most compelling challenges our interviewees also discussed this in detail(see Part III Bias).Rowes(2022)study of learners in a second-grade American classroom found that Google Translates programmin
78、g appeared to carry messages about what is considered appropriate and standard language use,disregarding nuances in language groups.One learner using the tool found that Tagalog was not listed as a language by Google Translate,and the only available option for the Tagalog-speaking pupil when transla
79、ting her own language to English was Filipino(which has been the official standardised language of the Philippines since 1987).Rowe(2022,p.884)reports that this left the learner in essence,engaged in a negotiation of what counts as a language,who decides what it is called,and which language was“corr
80、ect”.This suggests that by recognising some historical and political language boundaries over others,Google might re-enforce standardised language use.Artificial intelligence and English language teaching:Preparing for the future 18About the British CouncilThe British Council is the UKs internationa
81、l organisation for cultural relations and educational opportunities.We create friendly knowledge and understanding between the people of the UK and other countries.We do this by making a positive contribution to the UK and the countries we work with changing lives by creating opportunities,building
82、connections and engendering trust.We work with over 100 countries across the world in the fields of arts and culture,English language,education and civil society.Last year we reached over 75 million people directly and 758 million people overall including online,broadcasts and publications.Founded i
83、n 1934,we are a UK charity governed by Royal Charter and a UK public body.Implications for practice As English language learning is likely to be the most common discipline for AI use in education(Crompton&Burke,2023),English language teacher education and training must include a focus on AI literacy
84、.Teachers also need to develop their learners AI literacy so that they can understand the limitations and risks of AI and discuss the ethical issues around its use.Practitioners should carefully consider how models are chosen,as AI may carry messages about language use and exclude certain groups/var
85、ieties of English.AI can provide a conversational partner,provide language practice outside class and alleviate learner anxiety about speaking.However,more evidence is needed on whether the gains persist independent of such AI tools.Accessible and unambiguous ethics statements for AI in ELT should b
86、e developed and committed to,along with clear systems to ensure data privacy.Practitioners should be realistic about the current limited capabilities of AI and cautious about the hype.Image Mat WrightArtificial intelligence and English language teaching:Preparing for the future 19Areas for future re
87、search The majority of studies meeting our selection criteria were from Asia(72 per cent).Future studies conducted in a wider range of geographies would make results more widely generalisable.Many were conducted in higher education.More studies are needed in K12(school-level education)and adult lear
88、ning.The challenges of AI use in ELT are not as well reported as the benefits.Further efforts are needed to make explicit the challenges of using AI in ELT.Future research could focus on AI use for developing receptive skills,which did not emerge as strongly from the research data as productive skil
89、ls.Despite the rapid changes in available technology,conventional forms of pedagogy,such as lectures and explanations,persist.Future researchers could investigate and expand on how AI can create new opportunities for learning.There is a lack of research on specific tools and longer-term impact on le
90、arning,e.g.grammar,translation,AI-powered gaming.AI has clear applications for assessment,as well as implications for how learners skills can be assessed without them using AI for support.This is a significant area of need for research in future.Our systematic review provided a much-needed overview
91、of the field of AI and ELT,gathering the published research of the past decade.The findings reveal current benefits of using AI as well as some of the challenges and issues that need to be addressed.The British Council team were able to use these findings to help inform the questions and statements
92、for the teacher survey,the findings of which are presented in Part II,as well as the questions posed to the stakeholders interviewed for the third part of this report.Together,these three avenues of study help to triangulate the emerging themes and provide a more holistic understanding of the curren
93、t use of AI in ELT and how we might work to shape it for the future.Artificial intelligence and English language teaching:Preparing for the future 20Part IIThe survey:What teachers sayAbout the surveyThe perspective and views of English language teachers were underrepresented in studies analysed for
94、 the literature review.In response to this,a survey was designed to gather data on their use of and opinions about AI in ELT.This survey was shared through various ELT-focused social media channels and mailing lists,including the British Councils TeachingEnglish Facebook community.Contributing teach
95、ers Data was provided by 1,348 English language teachers,from 118 countries and regions,on where they teach,who they teach,how long they have taught for and whether they teach face-to-face,online or both.Respondents were mainly experienced English language teachers,with a majority(64 per cent)having
96、 taught English for over ten years and a further 21 per cent having taught for five to ten years.Only 16 per cent of respondents had taught English for fewer than five years.3 Just over half(53 per cent)taught exclusively face-to-face,while 42 per cent taught both face-to-face and online.Only six pe
97、r cent taught solely online.In terms of where they teach,state schools(33 per cent)were the most common,followed by private/fee-paying schools(23 per cent)and university/college(22 per cent).A smaller percentage(18 per cent)worked at English language schools,while 15 per cent chose Other(this includ
98、ed those working as private tutors,for in-company corporate training,volunteers or with NGOs and international organisations).Respondents taught a range of age groups,with 14-to 18-year-old learners most common(47 per cent),followed by 11-to 13-year-olds(36 per cent).Similar numbers taught universit
99、y(30 per cent)and adult(28 per cent)learners.Five-to ten-year-olds(20 per cent)and those under four years old(3 per cent)were the least common age groups taught.Respondents were from various world regions,with Asia(27 per cent)and Europe(27 per cent)being most represented,followed by the Middle East
100、 and North Africa(MENA)at 23 per cent.Eighteen per cent of respondents were from the Americas,while the percentage of respondents from sub-Saharan Africa(SSA)stood at four per cent.The survey:What teachers sayArtificial intelligence and English language teaching:Preparing for the future 22Figure 1 T
101、eacher survey respondent demographics3 Throughout this section,survey result numbers have been rounded,as a percentage,to the nearest whole number for readability.04 years58 yearsOver 10 yearsFace-to-faceOnlineBoth online and face-to-faceState schoolPrivate school (fee paying)University or collegeEn
102、glish language schoolOther(please specify)Ages 0-4Ages 5-10Ages 11-13Ages 14-18University studentsAdultsOther(please specify)MENASub-Saharan AfricaAsiaEuropeThe AmericasOceania1,348 teachersExperienceHow they teachWhere and who they teach118 countries and territoriesArtificial intelligence and Engli
103、sh language teaching:Preparing for the future 23How teachers are using AI in ELT The 1,348 respondents provided data on what AI tools they use and how they use them.A simple definition of AI headed this section of the survey:AI(artificial intelligence)refers to technologies that mimic human behaviou
104、r to conduct tasks normally done by people.Respondents were then asked to select from two separate lists:1)the AI-powered tools they used and 2)the specific tasks they used AI tools for(see Appendix A).Figure 2A Teacher survey results:Which AI-powered tools teachers useThe most popular AI-powered to
105、ols that respondents used were language learning apps(48 per cent).Language generation AI(37 per cent)and chatbots(31 per cent)were the next most widely employed.Automated grading(22 per cent),speech recognition software(21 per cent)and text-to-speech tools(19 per cent)had fewer reported users.Data
106、and learning analytics tools(12 per cent)and virtual and augmented reality tools(7 per cent)were the least used tools.A significant percentage of respondents(24 per cent)reported that they did not use any of the types of AI tools listed.AI-powered tools teachers useArtificial intelligence and Englis
107、h language teaching:Preparing for the future 24Figure 2B Teacher survey results:The tasks teachers use AI tools forFrom the list of specific tasks teachers used AI tools for(see Appendix A),the most selected options were creating materials(57 per cent)and helping learners practise using English(53 p
108、er cent).Creating lesson plans(43 per cent)and correcting learners English(33 per cent)were the next most popular selections.Grading or assessing learners(23 per cent)and admin tasks(19 per cent)were selected the least.However,18 per cent of the respondents said they did not use AI for any of these
109、purposes.What teachers use AI tools forArtificial intelligence and English language teaching:Preparing for the future 25Teachers views on AI in ELT The next section of the survey was completed by 1,112 teachers from 115 countries and regions.They were asked to rate 13 statements about AI in ELT(see
110、Appendix A)on a five-point scale of agreement:I strongly agree/I agree/neutral/I disagree/I strongly disagree.These statements reflected themes that had emerged from the review of the literature(see Part I),as well as other areas of interest.Some of the respondents also provided written comments(ran
111、ging from 84 to 204 per statement)to explain their rating decisions.The results are presented in this section.Analysis of the written responses to each statement was initially generated using the AI tool ChatGPT.These analyses were then reviewed,redrafted and added to by the publication authoring te
112、am.AI in ELT:the present4 Teachers responded to four separate statements in the survey itself.There were similar levels of agreement among the 1,112 respondents for all four statements,demonstrating that teachers saw little difference in the ability of AI to support learners development in these fou
113、r areas.Agreement that AI can help improve productive skills(76 per cent for speaking,75 per cent for writing)was similar to that for receptive skills(74 per cent for listening,79 per cent for reading).This is interesting because the literature(see Part I)suggests that in practice there is more focu
114、s on AI being used for productive skills,at least where research projects are taking place.In the 613 written explanations provided by teachers(204 for the statement on speaking,167 for writing,123 for listening and 119 for reading),several patterns emerged.Positive perceptions of AIs capabilities w
115、ere noted across these skills,with Figure 3 Responses from 1,112 teachers to survey statements 1 to 4 60.00%I strongly agreeAl can help learners improve their English speaking skills.Al can help learners improve their English writing skills.Al can help learners improve their English listening skills
116、.Al can help learners improve their English reading skills.I agreeNeutralI disagreeI strongly disagree50.00%40.00%30.00%20.00%10.00%0.00%Statements 1 to 4:AI can help learners improve their English speaking/writing/listening/reading skills.4Artificial intelligence and English language teaching:Prepa
117、ring for the future 26Figure 4 Responses from 1,112 teachers to survey statement 5 Statement 5:AI can have a negative impact on learners ability to improve their English.teachers praising innovative tools for learning,real-time editing,the ability to adapt to learners levels and offer engaging readi
118、ng materials.A recurring theme was AIs potential to enhance autonomous learning and provide non-judgmental practice opportunities.However,responses also pointed to the technologys limitations:AIs lack of human emotions,inability to fully grasp language nuances like humour,and concerns about over-rel
119、iance were cited.A shared belief across all four skills was that the integration of AI-powered tools and content should complement,rather than replace,existing methods.There was a range of responses to the question,indicating fairly balanced levels of agreement and disagreement among the 1,112 respo
120、ndents.The number of teachers who strongly agreed or agreed(36 per cent)was nearly equal to the number who strongly disagreed or disagreed(34 per cent).About a third of the respondents(30 per cent)felt more neutrally about this,which suggests some teachers have mixed feelings about whether AI might
121、have negative impacts or that its simply too early to make a judgement(perhaps because of a lack of experience with using AI-powered tools themselves).Many of the 129 written explanations provided expressed concerns about dependency,noting that learners might misuse AI or rely on it more than their
122、natural abilities.Quotes like Whats the point in learning English when AI can speak for me?and Students will rely too much on AI,resulting in a lack of confidence illustrate the perceived risk of over-reliance.However,several educators acknowledged that the outcome hinges on how the technology is us
123、ed and the quality of the AI product,suggesting the potential for both beneficial and detrimental effects.Guidance from the teacher was highlighted in some cases,it being pointed out that educators themselves need to be confident and have a clear purpose on how to use the tool for learning.That such
124、 issues are not exclusive to AI is also noted:Any teaching tool can have a negative impact if not used correctly.Common concerns about cheating,plagiarism and the potential for AI to replace critical thinking,while promoting a wooden,dead version of the language,were cited,indicating a cautious atti
125、tude towards AI in ELT.I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly disagreeArtificial intelligence and English language teaching:Preparing for the future 27Figure 5 Responses from 1,112 teachers to survey statement 6 The 1,112 responses showed general a
126、greement(70 per cent)with this statement,while only 12 per cent disagreed.A considerable number of the 120 written explanations expressed the belief that learners should be able to write independently to ensure skill development.Phrases such as students should develop their cognitive capacity and Im
127、 not sure AI-supported writing gives more than the illusion of progress in many cases highlight this perspective.Conversely,others acknowledge the potential benefits of AI tools for reinforcement and feedback,noting that they can help students learn and understand the writing process and that some s
128、tudents do seize this as a learning opportunity.The fear of over-reliance or misuse is again a common theme,with concerns about laziness or decreased creativity.However,the fact that such tools are now a common and accepted part of English writing is touched on,one respondent pointing out that if na
129、tive speakers use it why shouldnt learners use it too.Overall,while many respondents see value in AI-powered tools,there is a consensus that learners need to develop autonomous writing skills.Statement 6:Learners should be able to write in English without the help of AI tools(e.g.Grammarly,ChatGPT).
130、45.00%40.00%35.00%I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly disagreeArtificial intelligence and English language teaching:Preparing for the future 28Figure 6 Responses from 1,112 teachers to survey statement 7Statement 7:AI can plan effective English
131、language lessons for teachers.Responses highlighted broad agreement with this statement,with 65 per cent agreeing and only nine per cent disagreeing.The fact that a sizable 25 per cent remained neutral may suggest that this is a function of AI that is still less familiar to these teachers.A prevalen
132、t sentiment in the 143 written explanations is one of caution,with respondents asserting that while AI might offer structural or preliminary assistance,the human touch and ability to adapt plans whether from traditional sources or those generated by AI remains paramount.Statements such as AI is a to
133、ol,not a professional underscore the perceived superiority of expert human planning.However,others felt that AI was actually well suited to this task:Lesson plans are often formulaic,and AI excels in the formulaic.Some also acknowledged the potential of AI for saving time,generating initial framewor
134、ks to adapt and build on,and pointed out its ability to embellish existing plans.One teacher stated that,as someone new to teaching,their lesson plan quality has vastly improved due to AI.However,a recurrent theme was the need for teachers to review,check for errors and adapt plans for their learner
135、s,perhaps as they would with any provided lesson plan.Many respondents,either due to inexperience or scepticism,had not yet fully engaged with AI for planning.50.00%45.00%40.00%35.00%I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly disagreeArtificial intelli
136、gence and English language teaching:Preparing for the future 29Figure 7 Responses from 1,112 teachers to survey statement 8 There was predominant disagreement with this statement among the 1,112 respondents.Only 20 per cent felt sufficiently trained to use AI in teaching and 54 per cent felt inadequ
137、ately trained.A significant 27 per cent remained neutral,indicating possible uncertainty about their AI training needs.Disagreement with the statement from a majority of the respondents and analysis of the 100 written explanations show a predominant feeling of unpreparedness.A high number of respond
138、ents voiced that they had not received any formal training.One educator,from a private school,wondered if this lack of training is,in part,due to the fact that their private school is primarily selling access to human teachers and therefore does not believe AI-led education can be monetized.Another
139、pointed out that their initial training took place 16 years ago,making the lack of an AI focus unsurprising.Some educators have taken the initiative to self-educate,as noted in responses like I am trying to learn as much as I can by myself and I have completed online courses to develop my knowledge
140、of AI in teaching.A few indicated positive experiences,but the overarching theme is a call for more structured and comprehensive training opportunities.Statement 8:I have received enough training to incorporate AI into my teaching.40.00%35.00%I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00
141、%20.00%10.00%5.00%0.00%I strongly disagreeArtificial intelligence and English language teaching:Preparing for the future 30Figure 8 Responses from 1,112 teachers to survey statement 9Responses from 1,112 teachers indicate some disagreement with this statement.Only 23 per cent believed AI is more ben
142、eficial for English language teaching than other subjects.In contrast,42 per cent disagreed.A noteworthy 35 per cent remained neutral,suggesting possible ambivalence on the topic.A fraction of the 86 written explanations felt that AIs current proficiency was particularly apt for English teaching,wit
143、h comments such as AI excels at English language analysis and inference and students in EFL context need more practice outside the classroom.Conversely,many expressed the belief that AIs benefits spanned across all academic subjects,exemplified by remarks such as AI can be used for all subjects incl
144、uding science,commerce,history.There was also a substantial group of respondents who were uncertain or lacked enough knowledge to provide a definite stance.Overall,while some educators see AIs immediate applications in English teaching,a significant number believe in its broader pedagogical potentia
145、l.Statement 9:AI is more useful for English language teaching than other subjects.40.00%35.00%I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly disagreeArtificial intelligence and English language teaching:Preparing for the future 31Figure 9 Responses from 1,
146、112 teachers to survey statement 10 AI in ELT:the futureStatement 10:AI should be developed to support the learning of different varieties of English around the world(e.g.localised pronunciation and expressions).Responses indicated strong agreement.A substantial 80 per cent of the 1,112 teachers sur
147、veyed believed AI should be developed in this way,and only five per cent expressed disagreement.Many of the 84 written explanations expressed the value in this suggestion,noting Different people have their different language needs and that such an approach is more lifelike and natural.The importance
148、 of embracing diversity is clear in statements like If only some form of standard English is taught,this will reinforce many forms of bias and students need to be exposed to global Englishes.Conversely,some expressed reservations about diverging from standard English,as evidenced by comments includi
149、ng BBC English is enough and I think its not good cause speakers of English couldnt understand other speakers of English.Furthermore,there was recognition of the fact that this is not a new issue,or one specific to AI:Im not convinced that AI is the best way to achieve this,Software that promotes th
150、e use of language and cultural expression,within“videos”and/or animations,already exists.Overall,while many appreciated the need for more inclusivity and realism in language learning,concerns about standardisation and efficacy remain.60.00%50.00%40.00%30.00%20.00%10.00%0.00%I agreeNeutralI disagreeI
151、 strongly agreeI strongly disagreeArtificial intelligence and English language teaching:Preparing for the future 32Statement 11:By 2035,AI will be able to teach English without a teacher.Figure 10 Responses from 1,112 teachers to survey statement 11There was prevailing disagreement with this stateme
152、nt.Only 24 per cent of the 1,112 respondents believed that this would come to pass,compared to 51 per cent who expressed scepticism.However,26 per cent remained neutral,indicating significant uncertainty about the potential of AI in this area.A majority of the 125 written explanations expressed the
153、belief that while AI can aid the teaching process,it cannot substitute the unique human touch.Statements such as The human experience is unique and no machine can substitute it underscore the perceived value of human connection in education.A few,however,acknowledged technological advancements,sugge
154、sting that while AI may not replace teachers entirely,its role in education will grow:AI already knows more about language than most human teachers.There was also some acknowledgement of AIs potential to provide broader access to learning,with one respondent predicting that those who can afford it w
155、ill continue to prefer human teachers but the opportunities for those without means will expand exponentially.These 125 respondents generally appeared to think of teaching as something that will remain within a traditional classroom,a class of 20 to 40 kids AI wont be able to control the class.Overa
156、ll,while there was recognition of AIs potential,the consensus among those providing written explanations who are of course teachers themselves leaned towards the enduring importance of human educators.35.00%I strongly agreeI agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly di
157、sagreeArtificial intelligence and English language teaching:Preparing for the future 33Statement 12:AI and automated translation will eventually make learning languages unnecessary.Figure 11 Responses from 1,112 teachers to survey statement 12 Only 19 per cent of the 1,112 respondents agreed with th
158、is statement,while a sizeable 60 per cent disagreed.A notable 21 per cent remained neutral,indicating some uncertainty about the enduring value of language learning.Analysis of the 104 written explanations reveals a dominant scepticism towards complete replacement.Many respondents emphasised the dee
159、p cultural,social and emotional facets of language,as illustrated by comments like languages are beautiful and historical and let us know our roots and Learning languages is learning cultures,new ideas,different perspectives.Several educators recognised the benefits of improving technology but belie
160、ve the nuances of human interaction are irreplaceable,pointing out that AI will not make the human connection flow and that even minor miscommunications can lead to unforgivable misunderstandings among cultures with different traditions.The pleasure of learning languages was also raised by several r
161、espondents,one educator suggesting that Language learning may become slightly more niche or less widespread,but it will hopefully still be considered a valuable skill and some people love to learn a language purely for enjoyment.Some acknowledged the potential of AI in easing communication,especiall
162、y during short trips or visits,but overall the consensus underscores the enduring value of human interaction that doesnt pass through the filter of AI-driven translation.45.00%40.00%35.00%I agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly disagreeI strongly agreeArtificial in
163、telligence and English language teaching:Preparing for the future 34Statement 13:I worry about the impact AI will have on my role as an English language teacher.Figure 12 Responses from 1,112 teachers to survey statement 13 There was a balance of agreement and disagreement among the 1,112 respondent
164、s,with 38 per cent expressing concern by agreeing,36 per cent disagreeing and a significant 26 per cent remaining neutral.Analysis of the 84 written explanations provided reveals a predominantly optimistic outlook,with many educators believing that AI cannot replace them.Common views included the be
165、lief that All our kids(pupils)love us in a way they could never love working with AI only,indicating a strong trust in the unique human connection between teacher and learner.Some welcomed the additional tools and support provided by AI,believing that expert teachers who are early embracers of AI wi
166、ll adapt.Others were not concerned,for instance,one teacher said,Ill be retiring soon.However,concerns like I worry companies will want to abuse it to reduce staff costs were raised.Overall,while there is recognition of AIs potential,there was a prevailing view that teaching entails more than mere k
167、nowledge transfer;it involves human connection,emotional understanding and cultural appreciation,elements that respondents believe AI cannot replicate entirely.I agreeNeutralI disagree30.00%25.00%15.00%20.00%10.00%5.00%0.00%I strongly agreeI strongly disagreeArtificial intelligence and English langu
168、age teaching:Preparing for the future 35Key takeaways Teachers are using AI-powered tools for a range of ELT tasks.Teachers feel AI benefits the development of all four English language skills fairly equally.There is an even balance of teachers who see the potential for AI to impact on their learner
169、s English development either negatively or positively.Teachers generally feel that they have not received enough AI-specific training.There is an even balance between teachers who are concerned about the impact of AI on the teachers role and those who are not.Teachers lean towards the likelihood of
170、ELT remaining in the hands of human teachers.A significant number of neutral responses to several statements indicates a degree of uncertainty around AIs impact on the present and future of ELT.Summary AI-powered tools,or at least those labelled as such,are reported as being used by a significant nu
171、mber of the 1,348 English language teachers who responded to the first part of this survey.Several common themes emerged from the written explanations provided by some respondents.These expressed an optimism around AIs potential as a supplementary tool,notably its ability to provide tailored resourc
172、es,promote autonomous learning and improve specific linguistic skills.However,this enthusiasm is tempered by reservations about over-reliance on technology,concerns about diminished human interaction,the potential for misuse and the perceived inability of AI to truly grasp the subtleties of language
173、 and culture.There is a consistent emphasis on the irreplaceability of the unique human touch in teaching,highlighting the emotional,cultural and social facets of ELT.The lack of formal training and readiness for AI in the teaching space is again evident,with many of these educators feeling unprepar
174、ed and calling for better training in AI integration.In essence,while AI is seen as a promising tool in ELT,there is a strong consensus that it should complement rather than replace human-led teaching and learning.Artificial intelligence and English language teaching:Preparing for the future 36Part
175、IIIThe interviews:What our key witnesses say Approach Requests were sent to contacts around the globe,and those willing and available contributed their thoughts.The hope was for higher representation from outside the United Kingdom,but short timelines and the limited availability of interviewees mad
176、e this a challenge.Nevertheless,among the 19 interviewees,we were able to include voices representing 12 countries and territories.In terms of gender balance,there were fewer female contributors(five out of 19,or 26 per cent),but this reflects a ratio similar to estimates of the workforce in technol
177、ogy and STEM(science,technology,engineering and mathematics)more generally.5 Profiles of all the interviewees can be found in the Interviewee profiles section later in the report.To accommodate preferred styles of sharing ideas and peoples schedules,interviewees were given a choice of responding to
178、questions via a spoken interview or in writing.This has impacted the number and length of direct quotes used,as written responses are ready-made for quotation,whereas ideas within spoken interviews developed over longer exchanges.However,this has not affected the understanding of overall attitudes a
179、round the issues,and the conversations allowed for greater probing of ideas expressed by interviewees.The summary we present in the next sections aims to capture the key insights from contributors in a concise and accessible manner.The topic area clearly elicits strong reactions,and we hope we have
180、managed to do justice to the views and opinions expressed.The 11 themes we discuss here emerged from the authors collective analysis of the interview transcripts and submitted written responses to the interview questions.Naturally the themes will have been influenced by the questions asked(see Appen
181、dix B),and these questions,in part,were derived from trends and findings of note from our review of the literature(see Part I).The 11 themes,in no particular order or weighting,are AI and:definitions,pedagogy,Big Tech and neoliberalism,replacing humans,relevance for ELT,bias,teacher readiness,motiva
182、tion,inclusion,assessment,ethics frameworks and regulation.The interviews:What our key witnesses say This part of the report is based on the views of 19 stakeholders from across the globe:writers,academics,ministry of education representatives,chief executive officers(CEOs)of EdTech companies,traini
183、ng institute directors,teachers and teacher educators.The aim was to capture diverse perspectives in the discussion on AI in ELT and its role in the current scenario and future of our profession.The following text presents 11 major themes emerging from a series of in-depth interviews conducted with
184、these key witnesses.It was noteworthy from the systematic review(Part I)that the existing research studies on AI in ELT take place in a wide range of countries,many of which do not use English as their primary official language.This shows the far-reaching interest in,and importance of,ELT/L.For this
185、 reason,it was recognised as important to include stakeholders from as broad a range of geographies and contexts as possible.5 UK government census figures from July 2022 to June 2023 show that 26 per cent of the STEM workforce in the UK are female(Office of National Statistics,2023).Artificial inte
186、lligence and English language teaching:Preparing for the future 38Emerging themes DefinitionsSeveral interviewees began by highlighting the problem of definitions.AI is not a recent phenomenon(Roser,2022),and decades of AI evolution has meant the term covers very different types of technology:from m
187、achine learning algorithms to expert systems,personalised learning engines,large language models(LLM)and,more recently,to generative artificial intelligence(GenAI).These are very different technologies but are typically referred to with this same umbrella term AI.This lack of clarity has likely been
188、 exacerbated by a commercial imperative to cash in on the present enthusiasm for all things AI.Even tools that were previously described as simply digital or online are now being described as AI.Unsurprisingly,most EdTech suppliers will be looking to develop some form of AI enhancement for their non
189、-AI product.Interviewees did describe more nuanced definitions,for example the difference between strong AI(that can do any task a human can,and ultimately could be sentient)and weak or narrow AI(that performs specific,programmed tasks).What AI is and what it will be capable of is a contentious area
190、,and there is much discussion,as will be seen below.However,as a starting point,the lack of agreed definitions is not helping the debate.There is also possibly a need for domain-specific terms for AI in ELT that better communicate its uses in this specific context.When we talk about AI and we use wo
191、rds like generative AI or just AI generally,I think we do need to have a sense of history and a sense of which ones were talking about.Nicky Hockly,UK Many products that claim to be AI,are not AI.Al Kingsley,UK Key takeaways There is a clear need for a set of agreed definitions of Al so that when we
192、 discuss Al in ELT,we are talking about the same type of technology.We may require domain-specific definitions for Al in ELT.Artificial intelligence and English language teaching:Preparing for the future 39In my 35 years in the industry Ive yet to see a technology-driven language learning experience
193、 that is innovative,stimulating and makes use of modern pedagogical approaches.One look at the audio-visual methodologies employed by VR companies should be enough to show that we have a long way to go yet.Gavin Dudeney,UK There is a concern that learning may become individual,rather than individual
194、ised learning.In other words,AI should not become learners individually working to learn,rather it should promote cooperative learning,learning at their pace.Dr Ramanujam Meganathan,India PedagogySeveral interviewees spoke about AI and pedagogy,and this was usually unprompted or in response to more
195、general questions around how the technology might impact ELT.There was significant critique around less progressive theory that informed the instructional design of AI and EdTech more generally.The systematic review referenced in Part I of this publication finds evidence that substantiates this view
196、,with lecture emerging as a distinct pedagogical category in the research literature.Warshauer(1996)notes the influence of learning theory on the way that technology is used in education.For example,the early use of computers in language learning(computer-aided language learning or CALL)was influenc
197、ed by the predominant learning theories of that time i.e.behaviourism in the 1960s before moving to a more communicative approach in the 1970s and 1980s.As a result,in the early use of CALL,we find software with drills that require a user to repeat an utterance,or self-study quizzes with a single co
198、rrect answer,i.e.behaviourist/cognitivist.Similarly,Edgar(1995,p.1)sees parallels between the development of learning theory and the evolution of personal computer technology.The advent of personal computing allowed for constructivist approaches as individual learners were now able to experiment wit
199、h the computer as a tool in open-ended environments.This replaced the centralized and autocratic mainframes designed around behavioural objectives whose main purpose was content distribution.So,will AI allow for new pedagogies,or will we just see a new technology designed around existing ones?Where
200、interviewees commented on this,they expected the latter.Dr Ramanujam Meganathan,Professor of English at the National Council for Educational Research and Training,India,pointed out that,to date,AI has been mainly deployed to aid learning for the individual.This does not mean AI cannot also be used a
201、s part of a methodology that also prioritises collaboration,of course.The question is how best to balance the benefits of learners learning at their own pace with the aid of AI and the benefits of co-operative learning among peers.AI could be left to provide impactful self-study that then feeds into
202、 collaborative activity between learners.It could also take on an increasingly direct role in these collaborative elements of learning.Artificial intelligence and English language teaching:Preparing for the future 40The importance of a collaborative classroom came through strongly in many of the int
203、erviews,suggesting the need for methodologies and technologies that integrate AI into an ELT classroom where the focus is on learner-to-learner collaboration.Professor Rodney Jones,University of Reading,UK,was one interviewee to talk about a specific collaborative methodology,seeing a range of advan
204、tages that AI could bring to this type of approach.He spoke about the use of LLMs and generative AI to help learners develop their collaborative writing skills,while still working individually:There is something about assessment in universities that makes teachers reluctant to ask students to collab
205、orate on assignments.And students dont like to collaborate on assignments either.But the problem is that when they go out to the workplace,pretty much everything they write is collaborative and they dont know how to collaborate in their writing.And so,to have this kind of artificial classmate,this a
206、rtificial student with whom you can work to collaborate in your writing who is not going to be a free rider,whos not going to hurt your grade,whos not going to disappear when you have to hand in the assignment,but whos going to be responsible.Then thats enormously useful in at least creating a kind
207、of step for them to then learn to collaborate with human beings more in their writing.And of course,when they get to the workplace,theyll have to collaborate with both human beings and with machines.The always on,24/7 and potentially situated nature of AI was mentioned by several interviewees.This i
208、s not a pedagogy as such but the ability to access the learning you need at the time that you need it.This affordance could be described as anti-syllabus only learning what you require,i.e.organically,rather than learning superfluous items of a synthetic syllabus.However,it is worth noting that just
209、 in time and on demand learning is not a new phenomenon nor unique to AI.Advocates of mobile learning in the 2000s suggested this new capability,made possible by smartphones,would revolutionise education.While it is fair to say our lives have been transformed by mobile devices,formal state systems o
210、f education have not.At least,not yet.AI gives instant feedback,AI gives an instant response.Thats the teacher they dont have.So,the plus is there and 24/7.If I need help,desperately need help,Ill go to AI.Then the ball is rolling,then I can learn more and more.Dr Gumawang Jati,Indonesia Key takeawa
211、ys AI may have the potential to be transformative,but will it be held back with outdated learning theory?Using AI in collaborative methodologies has potential,but needs further exploration and research.Artificial intelligence and English language teaching:Preparing for the future 41So,there is perha
212、ps space in this global model for more local representation.It seems a bit of a shame to me that that local representation only occurs when the big ones like ChatGPT arent very good at it.But I think this is definitely a space.Nicky Hockly,UK Big Tech and neoliberalismBig Tech is a term that refers
213、to the most dominant or highly influential information technology companies.Concern around the influence that large,predominantly US-based,tech companies could have on ELT classrooms was evident,but perhaps not to the degree expected.A more nuanced understanding of the situation emerged with example
214、s where local companies and initiatives were responding to specific contextual requirements.Nicky Hockly,Director of Pedagogy at The Consultants-E,UK,for example,described a Berlin-based startup which,out of necessity,developed a machine translation product that worked with Ethiopian languages.Simil
215、arly,Joe Yiming Lee,a teacher and teacher trainer in Taiwan,described a project where authorities were proactively looking to create generative AI that was better suited to Taiwan,commenting that language generators like ChatGPT are not contextualised enough.This was linked to language,but also Asia
216、n knowledge,concepts and skills that were not properly understood or interpreted by AI trained on Western data.Others talked about commercial,neoliberal imperatives and the prioritisation of revenue streams over pedagogical soundness and validity.This is not an AI-specific characteristic:the global
217、EdTech market size was valued at US$106.46 billion in 2021,with North America having a 35 per cent share of that revenue(Yelenevych,2022).In line with this,interviewees still saw the US as the main player when considering Big Tech and AI,but there were several mentions of China,which has a lot of da
218、ta to work with and has been developing comparative technologies for at least as long as the United States.The perception is still that the US leads,whereas our review of the literature(Part I)shows a heavy skewing in favour of Asia-based research,overwhelmingly from China.Another important influenc
219、e was seen to be ELT publishers and decisions they will make,given the reach and impact that coursebooks still have.An important aside here was how AI would impact the creative process within publishing and the intellectual property rights of writers whose original work could be used by AI to produc
220、e new material for publication.There are obvious implications here(and not particularly positive ones)for authors,but also for how this plays out In the context of Africa,taking in all of the data that is coming from,if you wish,a Western perspective because most of the data is coming from there and
221、 then were applying it over here.And so,what does this say about our efforts?How does Africa come out of it?So yes,I think Im worried about that.Mohammed Mahmoud,Nigeria from a regulatory perspective.Dr Marcin Opacki,Assistant Professor,Institute of English Studies at the University of Warsaw,Poland
222、,said:I do not think that tech companies are a particular threat in terms of decision-making.I do however think that some decisions that can potentially be made by higher courts,such as the uncopyrightability of generated content or royalties for training data might change the scene in general.If ge
223、nerated content is rendered uncopyrightable,then using GPT content in ones work and presenting it as ones Artificial intelligence and English language teaching:Preparing for the future 42Researchers who publish articles about their AI developments and user testing generally focus on the technical de
224、signs and findings and hardly ever discuss social or cultural values,priorities or legacies.Sometimes they mention their countries policies and aspirations to be leaders in smart education and to transform their society.Professor Agnes Kukulska-Hulme,UK own will always amount to plagiarism from a le
225、gal standpoint.Whereas imposing royalty fees on training data content will have the potential to make it very difficult for tech companies to develop their software in the exponential pace that we are currently bearing witness to.Some interviewees also mentioned the step change brought about by the
226、recent pandemic and how this allowed tech companies to get into classrooms at a rate and scale not previously possible.This tied in to concerns about the dependence of national education systems on such technology and what happens when,for whatever reason,this becomes unavailable.Specific reference
227、to the collection of data by Big Tech did not come up as frequently as may have been anticipated.Also known as the Datafication of Education,this refers to the process whereby most of our everyday practices online and offline including aspects of the world not previously datafied and measured,such a
228、s social relations and emotions are converted into online quantified data,thus allowing for real-time tracking and predictive analysis(Van Dijck,2014,p.198).Professor Rodney Jones highlighted the dangers implicit in the race to endlessly improve AI by giving it what it needs most data,and every type
229、 of data:In order for it to simulate communication at that very,very high level,it is going to have to gather more data about us,yes.And so,it will,for example,have to be able to access our camera to be able to look at our embodied reactions.It will need to be able to access the location of our comp
230、uter.It will perhaps want to access all of our e-mails,our clicks and our Facebook friend page are all of these kinds of things in order to make it better and better.And we will lose sight of the dangers of that kind of mass surveillance if we focus just on this kind of imperative of making the AI a
231、 better and better teacher We may end up having a better teacher,but what are going to be the costs to our society,the cost of our privacy,and particularly our students privacy as well?Key takeaways It is not all Big Tech.There is both a place and a need for local,grassroots and more context-sensiti
232、ve AI.The US is still seen by some as the leader in this field,but our wider research shows that AI developments are happening around the world and particularly in Asia.Improving AI may require increasing datafication of our lives.Do we accept that?Artificial intelligence and English language teachi
233、ng:Preparing for the future 43AI simulations,regardless of their sophistication,cannot replicate the authenticity of human interactions While AI has made substantial progress in mimicking verbal human conversation,truly replicating both verbal and non-verbal cues in a manner indistinguishable from g
234、enuine human interaction remains almost impossible.Dr Nguyen Ngoc Vu,Viet Nam Replacing humansIn line with teachers canvassed in our survey,the majority of interviewees saw the future as one of continuity rather than widespread disruption of educational systems.This is because AI was seen to be able
235、 to do some of the things that humans do,but not everything that a good teacher does.These views could then be split into those that tended towards scepticism that any future AI technology would be capable of replicating all that is human,while others focused on the deficiencies of the current model
236、s that AI is based on.Dr Marcin Opacki said:To put it bluntly,I do not believe that any kind of AI is currently developed to a degree that would outclass human teachers.Even in light of remarkable technological progress,we should never underestimate the significance of the shared human experience or
237、 overestimate the potential of as I have said previously a Mathematical Optimization that stiches words together based on a predictive algorithm applied to contexts and topic areas.We can never dismiss the possibility that AI will supplant teachers one day,but it is decidedly not this particular typ
238、e of AI.A much greater technological leap would need to occur if this is to ever come to pass.Distinct human qualities such as experience,intuition,creativity and higher-level cognition came up quite frequently as the differentiators between AI and humans,and communication via natural language proce
239、ssing was only one part of the picture.Dr Marcin Opacki added:Any linguist worth his salt will tell you that as far as natural language is concerned,80 per cent of communication is extralinguistic,body language,gestures,context,etc.Meaning is therefore constructed through a complex interplay of pres
240、uppositions(shared knowledge),implicatures(intended meaning stated by the speaker),and inferences(how the recipient understands the message).A hypothetical fully immersive experience would need to not only model language accurately,but this complex interplay as well.Nevertheless,interviewees also sp
241、oke about where they did see AI replacing certain human teacher activity and,as detailed in our survey,teachers are already using AI for a range of tasks.Some interviewees suggested that teachers with AI,its not replacing the teacher,but replacing the types of tasks that take the teacher away from t
242、he students.Dr Gumawang Jati,Indonesia Artificial intelligence and English language teaching:Preparing for the future 44lower language proficiency and/or teaching experience would benefit the most from AI tools,a view supported by the comments of some surveyed teachers.This might be also backed up b
243、y research undertaken by Boston Consulting Group.The study(Candelon et al.,2023)found that consultants performing at the lower end of metrics were given a larger uplift from AI assistance in their work than those at the higher end.In terms of concerns around teacher quality and Sustainable Developme
244、nt Goal 4(SDG 4),6 this is certainly of interest.It could be that AI is able to give a larger professional lift to teachers more in need of help.Interviewees also touched upon learner preferences,mirroring a finding in the systematic review:learners may feel less anxiety around making mistakes with
245、an AI conversational partner than with a human one.Some commented that current AI may be better for conversational exchanges with lower-level language learners because such exchanges are more formulaic,have more predictable turn-taking and the context is less critical to understanding.Image InPress
246、PhotographyStudents like the remote AI to practise with so they can make mistakes by themselves,but need the human touch to build improvement and achievement.Wendy Edie,UK 6 SDG4:Ensure inclusive and equitable quality education and promote lifelong learning opportunities for all https:/sdgs.un.org/g
247、oals Artificial intelligence and English language teaching:Preparing for the future 45Professor Rodney Jones spoke about the limitations of using AI for spoken communication practice:One of the problems that many students have with communication is that they have this idea that conversation is this
248、kind of like a tennis game where you say something,I say something,you say something,I say something.And then when they actually get into a situation with the L2 and theyre actually having a conversation,they have a lot of trouble because conversation isnt like that.People talk over each other.There
249、s a lot of work that is done in managing things like turn taking and that sort of thing in conversation.And I dont think that AI is nearly there for that.And so,in that way,I think that probably better for lower proficiency learners in that regard rather than upper proficiency learners,who will prob
250、ably find conversing with an AI fairly easy but may have trouble conversing with real human beings because they manage conversation quite differently.Other interviewees pointed to the idea that AI is particularly suited to replacing the need for a human to do certain teacher tasks,for example writin
251、g a lesson plan.The survey of 1,348 teachers carried out for this publication(Part II)provides evidence that at least part of some teachers typical daily activity is being delegated to AI.For instance,one respondent to the teacher survey described AI as saving time and improving teaching quality,but
252、 that human analysis was still needed at the final stage:Lesson quality has vastly improved,as a new teacher,I have had to neglect adding to lesson plans,wasted hours doing dictionary searches I can do all research in seconds but yes sometimes not error free.Similarly,stakeholders interviewed voiced
253、 caution on the quality of AIs output,Thom Kiddle,Director,Norwich Institute for Language Education,UK,commenting:Ive seen experiments where you give it a reading text and ask it to make reading comprehension questions from it and its laughable what it does.You know its the 101 of item writing that
254、you would never fall in those traps.More research and analysis are required on which tasks could,should or should not be assisted by AI and where in the process a human still needs to be involved.This might take the form of an encyclopaedia or AZ of language teacher activity(inside and outside of th
255、e classroom),with a Key takeaways The majority view is that AI will not replace the need for human teachers any time soon and may never.AI is already aiding teachers with certain tasks.There is a need for more analysis of which teaching tasks can be done by AI and which should continue to be perform
256、ed by human teachers.breakdown of the advantages and disadvantages that AI can bring to each of those tasks.Any such resource would need to take the local context and teachers into consideration,e.g.low-resource/high-resource context,levels of digital literacy,teacher experience,knowledge.Despite re
257、cognition of its capabilities,there was almost universal agreement among the 19 respondents that AI is unlikely to completely replace human teachers.Interestingly,Dr Gumawang Jati,Senior English Lecturer,Institut Teknologi Bandung,and President of i-TELL Association,Indonesia,described an example wh
258、ere a costly model of sending expatriate teachers from China to Indonesia was being replaced with a combination of AI and online human monitoring.Still,even with this example,we see that human teachers are part of the process,albeit remotely.Only one interviewee,Professor Rodney Jones,had a differen
259、t perspective,stating it was inevitable that teachers would lose jobs:My most significant concern about generative AI in language teaching or in anything else is not the technology,but the fact that the technology is being driven by a kind of neoliberal imperative.And so,we talk very blithely about
260、AI,you know,being a kind of extra helper for teachers,being a kind of private tutor for students.But given that the imperative of governments and the imperative of businesses is going to be to cut costs,always will be to cut costs,I can guarantee you,I can guarantee you that teachers will lose jobs
261、because of that.Artificial intelligence and English language teaching:Preparing for the future 46Yes,AI is poised to have a significant impact on English language teaching and learning,possibly more than in some other disciplines.There is global demand for ELT.The massive demand means there is a con
262、tinuous search for efficient and cost-effective teaching methods,making AI-powered tools particularly attractive.Dr Nguyen Ngoc Vu,Viet Nam Relevance for ELTAn idea surfacing from our review of the literature(Part I)is that there is more research being conducted in ELT than in other subject areas.So
263、me interviewees suggested the level of demand for learning the English language is a reason for this.However,interviewees also referred to the impact of AI on other disciplines.There was discussion around whether LLMs are better suited to the hard sciences than social sciences and language because t
264、he former has such clearly defined concepts that are just easier to work with for what is basically a search and generation engine(Dr Marcin Opacki).Nevertheless,some observed that language learning is not a content subject(in the purest sense)and therefore would have different and perhaps more usef
265、ul applications of AI.An analysis by Reach Capital(Wan,2023)of EdTech GenAI tools in their pipeline presents language learning as a distinct category and second only to AI-powered study tools in the more general sense.Most interviewees saw enormous potential for the development of productive skills(
266、writing and speaking)and the immense benefit of AI feedback on this output came up frequently.The opportunity to practise speaking and get personalised,tailored feedback is a critical need in education systems globally.The constraints are well known:a lack of time within class for all learners to ge
267、t the speaking practice needed,teachers with low language proficiency struggling to provide a model for learners,teaching that treats English as a content subject rather than a skill,learners not being given the opportunity to speak due to teaching style,inability of a single teacher to provide feed
268、back to every single learner,the problems that come with large class size,and learner perception of the(lack of)usefulness of talking to a fellow learner who is not proficient in English.If AI could overcome many or all of these problems,it would bring significant change to ELT classrooms and,furthe
269、r down the line,English language learning outcomes.In contrast,input(i.e.reading and listening)was not seen to be as ripe for transformation via AI as output was.This marries with the findings from the review of the literature in Part I,where speaking and writing were prominent,and listening did not
270、 emerge as a focus.The use of technology to provide input is not new,and how AI might advance this was not a topic emerging from the interviews.The review of the literature also showed that AI that focused on the development of the skills of speaking and writing is not AI has the potential to signif
271、icantly improve the development of skills across a range of domains in the Ethiopian education system,including technical and vocational skills,industry-relevant skills,digital literacy and computational thinking.AI has the potential to meet changing needs in the labour market.Kedir Urji,Ethiopia Ar
272、tificial intelligence and English language teaching:Preparing for the future 47as prevalent.However,Professor Rodney Jones saw a focus on elements of the writing process as having some of the most exciting potential for AI:I think this is particularly important for academic writing in places like un
273、iversities because here again theres very,very little attention paid in British universities.I would say absolutely no attention at all to the process of writing,everything is about the product.You are assessed on the product.Youre never assessed on the process because actually theres no way for us
274、to know what the process is because theres no way that the process can be documented or can be recorded.But then if youre using generative AI as a collaborator,when you write,then that process is automatically being recorded.And as you have conversations with AI,as you critique the kinds of outputs
275、that it makes,as you refine the prompts that you give it to try to get it to create better outputs,this becomes the document of the process that youre going through.And so,I think thats also a really positive possibility for AI use in the teaching of writing.I see and hear instant feedback regularly
276、,as a benefit of learning using AI.You make a mistake and you get instant correction.There is,however,great benefit when improving fluency of delayed feedback,allowing learners to concentrate on communicating meaning rather than speaking with accuracy.A highly skilled teacher can gauge the usefulnes
277、s of immediate feedback vs delayed feedback on the spot.AI could be trained to behave the same way,but I dont see anyone doing it.Carla Wyburn,UK Key takeaways There is some evidence that AI will be more usefully deployed in ELT than in other disciplines,but,like the teachers in our survey(Part II),
278、not all interviewees were convinced by this idea.AI is seen as most useful for output,i.e.speaking and writing.AI may be able to help with teaching the process of writing,as opposed to simply focusing on the end product.Artificial intelligence and English language teaching:Preparing for the future 4
279、8Do we really want these biases carried over into what we teach and learn?Where are the gatekeepers?Gavin Dudeney,UK BiasThere was agreement among interviewees that bias is evident in AI.The issue of algorithmic fairness and biases has been noted in machine-learning research.Namely,bias prevalent in
280、 many societies based on gender,religion,ability,class,gender identity,sexual orientation and ethnicity may contribute to bias in AI in education systems(Ziesche&Kumar Bhagat,2022).Importantly,this bias may be harder for users to identify when it comes from computer systems that have a veneer of imp
281、artiality,as pointed out by Dr Maciej Rosiski,assistant professor at the Institute of English Studies,University of Warsaw,Poland:even though LLMs are neither representative linguistic corpora nor search engines,they are already treated that way by some students,teachers,journalists,and influencers,
282、etc.Users of LLM-based tools might be led to believe that an algorithm represents some kind of a consensus view on complex issues of all sorts(social,moral,scientific),when it does not have the capacity to do so.Labelling the algorithms as intelligent is a convincing frame in which answers generated
283、 by chatbots seem rational and objective,in contrast to the subjective decisions and opinions of individual people.Image Poyraz Ttnc/Pinhole IstanbulWhile there was concern across the board,some interviewees felt that because awareness of bias in AI was now quite high,this would elicit a response fr
284、om the tech industry and policy makers.Several referred to EU legislation(European Parliament,2023)in this area that will ban AI that is deemed as posing a threat to people,for example classifying people based on behaviour,socio-economic status or personal characteristics.This concern around AI and
285、the potentials for harm are reflected in the AI summit held in the UK in late 2023,which led to the signing of an international statement(the Bletchley declaration on AI safety),supported by more than 28 countries and the EU,that Artificial intelligence and English language teaching:Preparing for th
286、e future 49Does the system understand my cultural background in order to give me the very best answers based on my understanding of religion,my social upbringing and so on and so forth?So yes,its looking at that and trying to say,OK look can we have one that is talking to the African perspective?Moh
287、ammed Mahmoud,Nigeria acknowledged the need to address risks presented by advances in AI(Milmo&Stacey,2023).Others were less optimistic about an appropriate response from Big Tech.Dr Maciej Rosiski observed:Companies such as Microsoft,Google,or OpenAI,are not really that open when it comes to sharin
288、g information about the training data used for their language models.With so little transparency,its impossible to study their models scientifically,which I would expect before their tools get fully endorsed by education systems around the world.Its almost certain that the companies developing LLMs
289、do not care about regional variation in the English-speaking world,or other kinds of sociocultural variation.Relating specifically to ELT,our review of the literature(Part I)showed how AI can carry messages about appropriate and standard language use.Professor Rodney Jones talked about the difficult
290、y in challenging the bias when we do not fully understand why a system has behaved in a particular way:There is what computer scientists call the explainability problem with generative AI.We cant explain it.It works,but we cant explain how it works,cant explain why it works.And so,it becomes very,ve
291、ry difficult to question its results.And its only as good as the data that its trained on.If its trained only on samples of spoken English from majority English language speaking countries or countries where a lot of people are learning English,then you will have varieties of English or English pron
292、unciation that are not in its database,which may be perfectly comprehensible to others,but will be flagged as deficient because theyre not in the data set.Others pointed to the fact that when we talk about GenAI bias,we are typically referring to the output of Western technology such as ChatGPT.Howe
293、ver,there are many other non-Western technologies that may also exhibit bias,something that is arguably less discussed in Western media or academic papers.Will there be equal attention to bias in all AI systems globally?The challenge in addressing bias in AI becomes daunting when we begin to conside
294、r the number of systems,the number of versions of any one particular system and the fact that the track record in the regulation of technology has historically been poor.Refining the data used to train GenAI could be a way to positively affect outcomes in this case by removing bias.However,this appr
295、oach was questioned by some interviewees.Dr Marcin Opacki said:It is a misconception of sorts to think that the data need refinement in a prescriptive sense.In Large Language Models(again Foundation Models applied to natural language)any sort of alteration of the source data is in my opinion at best
296、 futile and at worst potentially detrimental to how the model works.The idea is for the model to make a prediction based on the actual frequencies and distributions typical to a given natural language corpus.The data simply need to be representative.There needs to be a lot of data and the data need
297、to be sufficiently varied.This might suggest that human flaws,including bias,are an inevitable part of LLMs while they exist in authentic human interaction(i.e.the corpus).Most contributors did not question attempts to address bias and other flaws through regulation or Artificial intelligence and En
298、glish language teaching:Preparing for the future 50refinement of training data.It was,however,seen by many as the teachers job to prepare their learners to critique AI:A critical look at machine-generated output can encourage students to take an ethical stance.I trust many will see why its not okay
299、to blindly trust an algorithm that cannot tell fact from fiction and is susceptible to all kinds of cultural biases.(Dr Marcin Opacki)Professor Rodney Jones took a similar stance in suggesting we re-evaluate our relationship with AI.It should never be viewed as a teacher(with all of the cultural exp
300、ectations and assumptions that go with that)but as a peer or a friend,and one that is flawed and generally not to be trusted:I think that if were telling our students this is your tutor,particularly in,you know,cultures where people are meant to have respect for their tutors.I think that sets up a r
301、eally counterproductive relationship with the AI Basically,if our students can see the AI not as a teacher but as a friend,and as like a really smart friend in many ways,but maybe a kind of friend with a lot of difficulties in socialisation who doesnt really understand much about how to talk to peop
302、le in a kind of polite or subtle way a kind of friend who can give feedback on the kinds of output that students are creating.Feedback which the students then have to interrogate and address with a critical eye.Key takeaways Bias is evident in Al and needs to be addressed.Regulatory frameworks can h
303、elp to manage bias from the top down,but these may be difficult to enforce universally.Teachers need to develop learners ability to critique their Al peer.Al should be set up not as a teacher but as a collaborator.Image Mat WrightArtificial intelligence and English language teaching:Preparing for th
304、e future 51Teacher readinessInterviewees felt that teachers are ill prepared for the increased use of AI.Moreover,existing teacher education and continuing professional development(CPD)provision was seen as lacking when it comes to developing teachers digital literacies.This poor track record to dat
305、e makes the challenge of preparing teachers for AI a sizeable one.An example given by Joe Yiming Lee showed how an educational system is responding to this challenge.This begins with general AI training for all teachers,followed by specific training for subject areas,recognising that the specificity
306、 of the discipline should also be taken into consideration.In India,Dr Ramanujam Meganathan noted that the new National Education Policy(NEP,2020)makes a case for use of AI as a mandatory element in both the initial and pre-service teacher education programmes.Thom Kiddle described how his teacher t
307、raining institute had planned to include a focus on AI in their teacher training courses in the summer of 2023,but found that two-thirds of the trainees had not even heard of AI,and that teachers had other priorities.In Indonesia,Dr Gumawang Jati spoke about a lack of expertise within the system bec
308、ause of the pace of change.One private-school teacher responding to our teacher survey(see Part II)observed the opposite,writing that although there was an in-school expert in place,they had received no AI training.This example points towards a conflict between AI and the schools business model,stat
309、ing Presumably,this is because they dont believe it can be monetized,as the school is selling access to human teachers.This friction between expectations of ELT and the incoming use of AI may also be reflected in the attitudes of some learners or their parents.Clearly,AI literacy is still a developi
310、ng teacher training area.For example,prompt design the textual instruction or input given to a language model like ChatGPT to perform a specific task was the only area within AI literacy that was mentioned across all of the interviews.As Nicky Hockly noted,there is a need to map out exactly what AI
311、literacy means in terms of specific,codified areas.Currently,this will need to be a rapidly evolving map,and revising content will be a near constant endeavour due to the fast pace of technological change.Nicky Hockly pointed out the ongoing change in the role of the English language teacher from a
312、general English teacher to more of a coach,noting,Theres There is one thing that concerns me and that is the kind of cheerleading that we see around Generative AI,which we saw around interactive whiteboards and so on.A very uncritical movement within our profession to take on board new technologies
313、without really thinking about them.Nicky Hockly,UK been discussion around this change of the role of the English language teacher for a while now towards the small facilitative coaching role because of all of the resources we have available online.Some interviewees were concerned by the lack of crit
314、ique around AI in ELT or of its blind acceptance as a good thing and a rush to integrate it.Many interviewees pointed to the huge number of AI talks at conferences,and that most of this focused on the practicalities of using AI rather than any critical discussion of the risks and how to mitigate the
315、m.AI,more than any other development in my time in the profession,has captured the imagination of teachers worldwide,but I am sceptical that many teachers understand how it works,and what its limitations are.Gavin Dudeney,UK Artificial intelligence and English language teaching:Preparing for the fut
316、ure 52Image Mat WrightIt is important that the teachers and parents are sensitised to help them understand AI as they are the first stakeholders that will handle learners.Dr Mugisha Annet Kajura,Uganda Key takeaways There is already a huge knowledge gap around digital literacies.Addressing Al litera
317、cy will be a massive challenge.Teachers have multiple competing priorities,and so teacher training in AI needs to be considered in the wider context of all the other demands on teachers.Education systems are beginning to grapple with this training need,but questions remain as to how they can keep pa
318、ce with the rate of change.Artificial intelligence and English language teaching:Preparing for the future 53MotivationA theme that emerged in interviews was the key role that motivation still plays in language learning,regardless of the technology deployed.Some interviewees noted how AI-powered tool
319、s had the potential to be inherently more motivating,but others felt that there was little evidence of this to date and what existed was mainly down to individual learner preference rather than a wholesale shift in the motivation dynamic.Dr Gumawang Jati observed that highly motivated learners will
320、get the most out of AI,the least motivated learners will ignore it and teachers that are interested in technology,they will play around with the AI,they will advise their learners on how to use it.Another aspect of motivation explored was how AI-powered automatic translation devices might negatively
321、 impact an individuals motivation to learn a language.If a technology could do it for you,why would you invest the time in learning a language at all?Here there was both agreement and push back,with interviewees providing a variety of drivers for learning a language even after the improvement of tra
322、nslation tools.Al Kingsley,Multi Academy Trust Chair and EdTech CEO,UK,did agree(sadly,yes),but also noted that for gaining and maintaining employment,competency in a language may become even more of a differentiator than at present.In a competitive marketplace,the ability to actually speak a langua
323、ge may become an even more sought-after skill.Some interviewees also pointed out there will remain those learners that have no reason or motivation other than simply the love of language and tackling the immense but enjoyable task of learning one.Lastly,an interesting point about the potential of au
324、tomatic translation was how it might be a force for including the linguistically disenfranchised.Dr Marcin Opacki commented:The good thing about this is that this has the potential to help a lot of people,who have thus far been in a state of cross-cultural exclusion,partake in the bounty of globaliz
325、ation.We especially in the community that works with or around foreign languages,meaning linguists,teachers,publishers,course designers,etc.often forget that there are scores of people who have never been successful in learning a foreign language and never will be.Why not finally give them some agen
326、cy when travelling or interacting with foreigners?Language is not just to communicate.Its a way to develop an understanding of the history and culture of those who speak it.Speaking a foreign language enables people with different backgrounds to form strong personal connections.Everyone speaking the
327、ir own language and using AI to understand each other may lead to a more efficient world,but also one that is far less colourful,far less interesting and with far more insular thinking.Carla Wyburn,UK Key takeaways Motivation remains a barrier or enabler to learning.AI does not appear to be changing
328、 that,yet.There are many reasons for learning a language,and automatic translation tools will not mean the end of language learning,a(very)human activity.Automatic translation tools could promote inclusion by aiding the less linguistically confident.Artificial intelligence and English language teach
329、ing:Preparing for the future 54InclusionSome interviewees believed that AI has the potential to improve accessibility for some learners.For example,being able to use the spoken word to interface with a computer will significantly aid those that are unable to use a keyboard or who are visually impair
330、ed.Generally,assistive technology enhanced by AI was seen to have great potential,albeit not yet realised to a great extent.One interviewee questioned whether AI for accessibility would itself receive the amount of attention and investment required,being more likely to see incidental advances from m
331、ore general technological development(for example,speech recognition was not developed with disability in mind).Al Kingsley also spoke about learners who might be excluded from formal schooling due to a range of reasons,such as social anxiety,and how they might benefit from AI-powered educational te
332、chnology.In terms of equity of access and the digital divide,interviewees expressed less positivity.Generally,AI was not seen to be worsening the digital divide per se,but it was also not seen as part of the solution.There is some existing scepticism around technology and learning,with its use in sc
333、hools generally not The AI tech companies cater to the rich and elite schools and families.The poor people and government schools do not have much access to AI and materials.Dr Ramanujam Meganathan,India Image Mat WrightConcerns arise that a lack of proficiency in using AI-powered tools might widen the gap in learning opportunities.Dr Toshiyuki Kanamaru,Japan Artificial intelligence and English la