[Neenaz Ichaporia] Hello, and welcome everyone.
[Neenaz Ichaporia] Thank you so much for joining us for the great AI debate today at our Elton's Festival of Innovation.
[Neenaz Ichaporia] I'm Nina Suchaporia, ed tech lead at the British Council, and I'll be chairing this session.
[Neenaz Ichaporia] So today, we're discussing a topic of concern for educators
[Neenaz Ichaporia] worldwide,
[Neenaz Ichaporia] AI.
[Neenaz Ichaporia] And more specifically, we will tackle the question,
[Neenaz Ichaporia] is AI a force for good in language education?
[Neenaz Ichaporia] Joining me for the debate today are a stellar panel who are all experts in their fields, and I'm very excited to introduce them to you.
[Neenaz Ichaporia] First up, we have Elena Onchevska
[Neenaz Ichaporia] Aga, who is full professor in applied linguistics at Saint Cyril and Methodius University in Skopje, North Macedonia.
[Neenaz Ichaporia] Her research interests revolve around language teacher education
[Neenaz Ichaporia] with a focus on quality teacher mentoring and helping it become widely available.
[Neenaz Ichaporia] Elena is also cofounder of noticing,
[Neenaz Ichaporia] an AI backed web based facilitator of mentoring practices.
[Neenaz Ichaporia] She's particularly interested
[Neenaz Ichaporia] the ways in which humans can meaningfully collaborate with AI, what is known as hybrid intelligence or HI.
[Neenaz Ichaporia] We also have with us today, doctor Mariano Felice.
[Neenaz Ichaporia] He is a senior researcher and data scientist for language assessment and learning at the British Council.
[Neenaz Ichaporia] Mariano's role involves researching the application of natural language processing to language assessment,
[Neenaz Ichaporia] as well as providing strategic guidance for the development and adoption of AI products and promoting AI literacy.
[Neenaz Ichaporia] Our third panelist is Guy Zaslavsky
[Neenaz Ichaporia] from Plingo.
[Neenaz Ichaporia] Guy has been leading EdTech Ventures for over 15 years.
[Neenaz Ichaporia] And notably, he's led on developing market leading EdTech and language learning platforms, including the Bloomberg exam prep line of adaptive online courses.
[Neenaz Ichaporia] And finally, we're very pleased to welcome
[Neenaz Ichaporia] Neeti Tripathi.
[Neenaz Ichaporia] Neeti is an experienced educator and principal
[Neenaz Ichaporia] leading innovation in international education across international baccalaureate,
[Neenaz Ichaporia] Cambridge, and Indian syllabi
[Neenaz Ichaporia] with nearly 20 years experience. So thank you so much panel for joining us, and thank you so much to our audience. I can see you're joining us from all over the world. It's truly an international audience today.
[Neenaz Ichaporia] First, I'd like to kick off by asking you, the audience, to share your views.
[Neenaz Ichaporia] Our first poll should appear on your screen in a moment.
[Neenaz Ichaporia] So what we'd like to ask you today is who here is already using AI
[Neenaz Ichaporia] to support their language teaching and learning?
[Neenaz Ichaporia] So we should have the results coming in now.
[Neenaz Ichaporia] And you have a choice of, uh, of options to choose from. You have yes all the time, sometimes,
[Neenaz Ichaporia] only rarely, or no, not at all.
[Neenaz Ichaporia] Thank you so much audience for voting.
[Neenaz Ichaporia] Um, and I think we can bring up those audience results now. Um, there you go. We have the results. So
[Neenaz Ichaporia] what well, a majority of you, almost 70% of you are using AI to support language teaching and learning. At least
[Neenaz Ichaporia] sometimes,
[Neenaz Ichaporia] some of you are using it all of the time.
[Neenaz Ichaporia] And, um, another about 20, 26%
[Neenaz Ichaporia] said only rarely or not at all. So most of you are using AI, and today's debate is going to be really, really relevant for all of you who are either using AI or curious about how to use AI in language education.
[Neenaz Ichaporia] So in today's debate,
[Neenaz Ichaporia] um, I will ask the panel members a number of questions about different aspects of AI in language education. And at the end of the debate, we'll invite you, the audience, to share your views again through the poll. So please do participate and have your say. If at any point you have any technical difficulties, the Elton's team are here to help you via the chat. And if you have any questions for our speakers,
[Neenaz Ichaporia] please type them in the q and a section at the bottom of your screen.
[Neenaz Ichaporia] So, um, let's go to the panel now. Let's begin with the basics.
[Neenaz Ichaporia] Let's start with what is AI and what are some examples of AI in a language learning or teaching context. And, Mariano, I'd like to come to you on this first. Could you give us some examples of AI and briefly explain to us what really is AI in the context of language teaching and learning?
[Mariano Felice] Oh, yeah. Hi, Nina. Hi hi, everyone.
[Mariano Felice] So, um, well, nowadays, there are lots of different definitions of AI, especially as the as the lines are becoming
[Mariano Felice] increasingly blurrier.
[Mariano Felice] But I think in simple terms, we can define AI as the ability of computers to perform tasks that typically require human intelligence.
[Mariano Felice] And what we mean by this is tasks that
[Mariano Felice] require learning, reasoning,
[Mariano Felice] uh, problem solving, understanding language,
[Mariano Felice] and so on. So these are generally
[Mariano Felice] complex tasks.
[Mariano Felice] And within AI, we have different subfields.
[Mariano Felice] So we have things such as natural language processing, so the subfield that deals with everything related to language, understanding and producing language. We also have speech recognition, for example,
[Mariano Felice] uh, that deals with audio
[Mariano Felice] and things like computer vision where we can
[Mariano Felice] generate and process,
[Mariano Felice] um, images and video and so on. And in the context of education
[Mariano Felice] and language planning in particular, I think we've seen many examples. Many of us are already probably very familiar
[Mariano Felice] with some of these tools. So I think we've seen things like chatbots, for example, for conversational practice.
[Mariano Felice] Uh, we have tools for grammatical error corrections, so to spot grammar mistakes
[Mariano Felice] and help you with expressing yourself in English,
[Mariano Felice] um, much better.
[Mariano Felice] And then, uh, for practicing speaking skills, for example, you can use technologies such as speech recognition or speech to text.
[Mariano Felice] I think many of us have also seen examples of automated scoring, for example, in in tests
[Mariano Felice] and also things like feedback that we get, for example, using tools such as LLMs
[Mariano Felice] and generative AI. So I think those would be probably the most familiar examples that that we've all seen and know.
[Neenaz Ichaporia] Thanks a lot, Mariano, for those insights. Um, and there is a audience member who's added to what you said as well. Um, Iqbal says, um, also generating activities and games to use during classes,
[Neenaz Ichaporia] and, um, that reflects some recent research the British Council conducted into AI and ELT, and, actually, um, a lot of teachers told us that they were using it for generating lesson materials and lesson plans. Thank you so much, Mariano, for sharing those insights.
[Neenaz Ichaporia] Um, I'd like to turn to Elena now. Elena, from your experience, what aspects of language education, both learning and teaching, can AI help with and how?
[Elena Oncevska Ager] Yes. Thank you. Um, when it comes to supporting learning, be it in the context of
[Elena Oncevska Ager] learning languages or learning teaching in my world we're all learners
[Elena Oncevska Ager] alongside our students. Um AI can offer can offer up a vast range of personalized
[Elena Oncevska Ager] immersive and safe environments
[Elena Oncevska Ager] for learners to
[Elena Oncevska Ager] explore things, to choose their own learning paths if you like
[Elena Oncevska Ager] and their preferred ways of learning. And we know from the literature that these these are prerequisites for involved enjoyable lasting learning. We'll we'll learn through active involvement,
[Elena Oncevska Ager] through powerful experience,
[Elena Oncevska Ager] by exercising our agency in building knowledge.
[Elena Oncevska Ager] And this kind of responsibility, I feel, is central to to learning.
[Elena Oncevska Ager] Now by,
[Elena Oncevska Ager] uh, having AI generate for us, I think,
[Elena Oncevska Ager] um, we run the risk of, uh, these knowledge building processing
[Elena Oncevska Ager] processes being narrowed down a little, positioning us at times as spectators of knowledge building, not active agents.
[Elena Oncevska Ager] So instead of having AI
[Elena Oncevska Ager] generating content for us, I would like to see
[Elena Oncevska Ager] AI
[Elena Oncevska Ager] more used for
[Elena Oncevska Ager] generating better, richer content, you know, helping us generate
[Elena Oncevska Ager] better, richer content of our own. This is exactly what we try to do at noticing network. We we try to help teachers reflect, plan,
[Elena Oncevska Ager] investigate their classrooms through dialogue with AI models.
[Elena Oncevska Ager] These dialogues are designed to prompt richer thinking.
[Elena Oncevska Ager] Um, and I think, like, in any good pedagogy,
[Elena Oncevska Ager] um,
[Elena Oncevska Ager] when using AI, the learner needs to remain center stage,
[Elena Oncevska Ager] uh, with the teacher or the mentor or the AI model facilitating this knowledge building
[Elena Oncevska Ager] for the learner
[Elena Oncevska Ager] and not doing the knowledge building for them. The result is better preparedness for learning language in new ways for students or for
[Elena Oncevska Ager] teachers to feel better prepared to design more engaging
[Elena Oncevska Ager] learning
[Neenaz Ichaporia] experiences. Thank you very much, Elena. And I think, um, what you're saying is also being echoed by our our audience in the chat. So we have a question from
[Neenaz Ichaporia] Wasila.
[Neenaz Ichaporia] Please forgive me if I've mispronounced your name. But the question is using AI for teachers might be good and useful, but what about learners? They're using AI instead of using their brains. Um, I know that this is a concern that many educators share about about an overreliance on AI or, as you said,
[Neenaz Ichaporia] um, AI doing the task instead of us doing the task. Mhmm. Um, so thank you so much for sharing those insights. I'd like to, um, I'd like to get the other panelist view on this as well. Kai, would you like to come in on and comment on what are some of the areas in language teaching and learning where we can use AI beneficially?
[Guy Zaslavsky] Sure. And, uh, just about this comment this participant made,
[Guy Zaslavsky] when using AI is not necessarily maybe a lot of students think about it. I'm going to use AI to help me with my homework.
[Guy Zaslavsky] But, uh, we primarily see it as as using the AI to,
[Guy Zaslavsky] um,
[Guy Zaslavsky] either supplement
[Guy Zaslavsky] or, in some cases, replace your teacher or
[Guy Zaslavsky] replace your standard home boring homework.
[Guy Zaslavsky] Um, in Plingo, the the app we're developing for English learners, uh, we're using AI extensively.
[Guy Zaslavsky] Mhmm.
[Guy Zaslavsky] And,
[Guy Zaslavsky] um, generative AI is used, uh, for creating basically lots of, um,
[Guy Zaslavsky] really engaging content. So we think one of the challenges today
[Guy Zaslavsky] is that,
[Guy Zaslavsky] um, if you just wanna teach, let's say, some
[Guy Zaslavsky] vocabulary, some nouns, then it's very easy to just pull out a picture of penguin or
[Guy Zaslavsky] or, you know, ice cream. But then if you wanna make a memorable immersive experience,
[Guy Zaslavsky] you could have generative value to create create for even a video of a penguin eating ice cream, and then you involve it in some kind of interactive task.
[Guy Zaslavsky] You create a much more memorable,
[Guy Zaslavsky] uh, engaging experience that has more learning value.
[Guy Zaslavsky] Also, speech recognition,
[Guy Zaslavsky] I think, is is more than, uh,
[Guy Zaslavsky] more
[Guy Zaslavsky] interesting in the context of language learning because
[Guy Zaslavsky] a lot of, uh, language learners
[Guy Zaslavsky] do not
[Guy Zaslavsky] uh,
[Guy Zaslavsky] use language in the same way as most speech recognition systems
[Guy Zaslavsky] nowadays.
[Guy Zaslavsky] Powered by AI have been trained for us. Typically, you would have English speakers,
[Guy Zaslavsky] native English speakers used as the training data for regular
[Guy Zaslavsky] speech recognition models, uh, one of the things we had to struggle with, and we came to pretty good results, is recognizing
[Guy Zaslavsky] speech by learners where they have a lot of pronunciation errors,
[Guy Zaslavsky] maybe very strong accents.
[Guy Zaslavsky] In some cultures, you know, Japanese, for example, have even entire phonemes that they're almost incapable of pronouncing initially.
[Guy Zaslavsky] So that makes it a very interesting AI problem.
[Guy Zaslavsky] There are also AI, uh, opportunities when it comes to, uh, assessing mastery
[Guy Zaslavsky] and predicting performance. So if you want to always
[Guy Zaslavsky] cater to the student,
[Guy Zaslavsky] uh, learning activities
[Guy Zaslavsky] or tasks that are not too difficult, not too easy,
[Guy Zaslavsky] and take into consideration all of their past learning history. You would probably need a nice AI system for that.
[Guy Zaslavsky] Um, and, um,
[Guy Zaslavsky] I would say there's also meaning interest, understanding the meaning behind their their inputs as part of a early development of conversational skills,
[Guy Zaslavsky] which we think is really, um, vital.
[Guy Zaslavsky] Lastly, I would say pronunciation training is what, uh, AI can offer unique benefits in.
[Guy Zaslavsky] So it's not just speech recognition, but actually coaching individual
[Guy Zaslavsky] users,
[Guy Zaslavsky] uh, on improving
[Guy Zaslavsky] both individual phoneme pronunciation
[Guy Zaslavsky] and entire words
[Guy Zaslavsky] and and
[Guy Zaslavsky] the flow of sentences,
[Guy Zaslavsky] I think this can also be a place where it'll be a game changer.
[Neenaz Ichaporia] Thank you very much, Guy. And you said so many very interesting things there. Um, I think one of the things that,
[Neenaz Ichaporia] resonated with me is that pronunciation is an area that in our research we found,
[Neenaz Ichaporia] um, there's a lot of work being done, as you mentioned, around helping students correct their pronunciation using, um, AI tools.
[Neenaz Ichaporia] Adriana in the chat has also commented about how some AI tools can be useful for differentiated instruction, and I think that's partially what you were saying along with personalized learning as well. So thank you so much for sharing those perspectives. And,
[Neenaz Ichaporia] Niti, I'd like to come to you now. As a principal of a school, what sort of benefits are you seeing when it comes to using AI in education?
[Neeti Tripathi] Yeah. Uh, thank you for the question, and hello, everyone.
[Neeti Tripathi] Uh, well, I see lots of benefits. Uh, first of all, uh, Elena also spoke about the personalized learning.
[Neeti Tripathi] So it's a really good tool which can help the teachers and the school leaders to
[Neeti Tripathi] make sure that each and every child because we have kids of all sort of abilities, child needs,
[Neeti Tripathi] wants. So AI definitely can help us to go and reach to each and every child. So personalized learning pathways are available
[Neeti Tripathi] with the help of AI. Then another thing is that it can help us, the educators, the leaders, the teachers,
[Neeti Tripathi] provide effective assessment to plan effective assessment and also give real time instant feedback.
[Neeti Tripathi] So you know the learning and whether learning is happening or not can be mapped together and can be monitored at the same time. Many times we see the teaching is happening or learning is not happening, so we can bridge that gap with the help of AI.
[Neeti Tripathi] Then I feel as a as leader of the school, I feel that with the help of AI, we can take a lot of data driven evidence based decision
[Neeti Tripathi] where we have to shape our school's future policies or pedagogies
[Neeti Tripathi] and have to be coming up with new technologies, involvement in our schools. So AI can definitely help us to take those database decisions.
[Neeti Tripathi] I also feel that AI is a great assistance to teacher. It save a lot of time. They can definitely,
[Neeti Tripathi] uh, you know, uh, take AI's help, and it can
[Neeti Tripathi] definitely
[Neeti Tripathi] help them to enhance their efficacy,
[Neeti Tripathi] their efficiency because, obviously, it is here it is there to assist them.
[Neeti Tripathi] Then I feel, again, we have diverse learners.
[Neeti Tripathi] Uh, it can help us to take, uh, to cater to their need also.
[Neeti Tripathi] Another thing which I feel that nowadays AI is really being used to
[Neeti Tripathi] plan and execute engaging learning experiences in the class so that, you know, often the kids do not get they don't get bored. They get involved. They get motivated.
[Neeti Tripathi] They want to participate.
[Neeti Tripathi] So AI can definitely help us in those areas.
[Neeti Tripathi] Another with another thing which I really feel is very good, again, as a school leader that obviously with new AI tools coming up, we have to upgrade ourself.
[Neeti Tripathi] So teachers can use it as a tool to have their CPD, continuous professional development.
[Neeti Tripathi] Educators, leaders can plan because definitely we need to learn
[Neeti Tripathi] about AI. We need to work with AI, and we have to, uh, know about AI. So we have to upgrade ourselves. So a lot of PD sessions can plan keeping that in mind.
[Neeti Tripathi] Well, another thing is we are talking a lot about ethical use of AI. So, again, it helps and it forces, I would say, the schools to think into that direction also that we come up with some kind of AI using policy in our school. We think that how ethical concerns are, uh, being taken care of. So, uh, we are not closing our eyes from that side that somebody asked in the chat. How about the learner? They use it, you know, for their homework. Yeah. So we have to make sure there's a good balance between that. So we've got to work towards it.
[Neeti Tripathi] I also feel that, you know, with the use of AI, we can definitely take care of the parental engagement in the school. There is a time when it really becomes a school for difficult to get into touch with all the parents and listening to their issues, but maybe AI can definitely help us in that area also.
[Neeti Tripathi] And last but not the least, I feel that, uh, we talk a lot about fostering 21st century school, uh, skills nowadays, collaboration, communication,
[Neeti Tripathi] decision making, you know, presentation. So definitely, I can add a lot into that area also. So these are some benefits I see while using AI into schools
[Neeti Tripathi] system and having its impact on education.
[Neenaz Ichaporia] Thank you so much, Niti, for sharing your experience, um, based on, um, your your work as head of the school.
[Neenaz Ichaporia] Um, there's an interesting comment from Chris that links into what you were saying in the chat. He says, I don't want Jenny. I have to generate materials for me, and I don't want my students to generate materials to send me. I want Jenny. I have to provide materials based on students' recordings and writing that will help them learn more English.
[Neenaz Ichaporia] So, again, it's interesting to see all the teachers and the educators in the chat thinking about and really expressing what they would like AI to help them with.
[Neenaz Ichaporia] Um, there are, of course, a lot of claims around what AI can do. Um, let's think more about the evidence behind these claims. So, Elena, I'd like to ask you this next question as a practitioner, but also as an academic.
[Neenaz Ichaporia] What evidence is there for the efficacy of AI powered learning solutions, and what evidence should we be looking for?
[Elena Oncevska Ager] Um, well, the evidence I'm looking for whenever I review any use of AI is whether it supports learner agency.
[Elena Oncevska Ager] By learner, I mean, student agency, teacher agency. I think we're all learners as I said previously.
[Elena Oncevska Ager] So does it support agency?
[Elena Oncevska Ager] Um, does it rely on the learner's active involvement?
[Elena Oncevska Ager] Uh, do, uh, does the learner feel better prepared
[Elena Oncevska Ager] for their engagement with the AI?
[Elena Oncevska Ager] Um, there is, of course, increasing evidence of AI successfully supporting learners in, um, developing their productive skills, mostly speaking, writing.
[Elena Oncevska Ager] AI can act as a conversation partner to help with pronunciation,
[Elena Oncevska Ager] accuracy, fluency when it comes to speaking, while also, uh, as research suggests, um, alleviating
[Elena Oncevska Ager] speaking anxiety.
[Elena Oncevska Ager] AI can support the development of writing by checking learners, grammar, and vocabulary, acting as a companion
[Elena Oncevska Ager] writing companion if you like, as well as providing feedback on their writing. Though I'm a bit skeptical about the, uh, this last bit, the providing feedback on writing because our experience so far suggests that AI models
[Elena Oncevska Ager] may not be as honest
[Elena Oncevska Ager] as we might want them to be because they're trained to please
[Elena Oncevska Ager] and research
[Elena Oncevska Ager] sometimes takes precedence.
[Elena Oncevska Ager] As for teachers, um, AI for lesson planning appears to be rather popular though what we mean by this can range from,
[Elena Oncevska Ager] um, I don't know, fill in a quick form and get
[Elena Oncevska Ager] an 80% finished lesson plan in minutes
[Elena Oncevska Ager] to what we do at noticing network
[Elena Oncevska Ager] which is to to simulate a pre lesson mentoring discussion,
[Elena Oncevska Ager] um a dialogue of sorts which is meant to enrich teachers thinking by getting them to revise things, notice any inconsistencies
[Elena Oncevska Ager] or incompleteness in their thinking,
[Elena Oncevska Ager] um, or or find those important moments of confusion which which trigger new thinking.
[Elena Oncevska Ager] Um, the the the idea is to help them
[Elena Oncevska Ager] get better prepared for the challenges
[Elena Oncevska Ager] of the classroom not by any means do any of the lesson planning for them. I think I'm very precious about lesson planning.
[Elena Oncevska Ager] Um there's also some evidence that uh AI can help
[Elena Oncevska Ager] in the context of AI supported collaborative learning. For instance,
[Elena Oncevska Ager] um, using AI as
[Elena Oncevska Ager] an additional discussion like Zoom do,
[Guy Zaslavsky] which can enrich conversations while role modeling productive group engagement for
[Elena Oncevska Ager] learners who might not be
[Elena Oncevska Ager] well versed in group work.
[Elena Oncevska Ager] Uh finally AI now makes it possible for teachers to design adaptive gamified
[Elena Oncevska Ager] environments
[Elena Oncevska Ager] uh, which can lead to yet more immersive learning. I'm very excited about this because I'm all for, you know, learners having memorable memorable experiences aligned to their to their needs, um, along
[Elena Oncevska Ager] along the lines of what Niti was talking about.
[Neenaz Ichaporia] Okay. That's that's a fantastic
[Neenaz Ichaporia] sort of rundown of all of the sort of evidence based uses for AI in language teaching and learning. Thank you so much, Elena.
[Neenaz Ichaporia] Um, I wanted to share,
[Neenaz Ichaporia] um,
[Neenaz Ichaporia] as you said, evidence. Evidence is so important. And you're looking for evidence of does it support learning agency and how what is the learner's experience?
[Neenaz Ichaporia] Tatiana's left a very interesting
[Neenaz Ichaporia] comment in the chat, or she shared an experience where she says we used an AI app for pronunciation training of adult students, and it was based on American English. So our native British teacher scored rather low. Um, it's just an interesting anecdote anecdote, I think, that shows what happens when maybe it doesn't work the way you expected it to.
[Neenaz Ichaporia] Mariano, I'd like to ask you the same question I I asked Elena.
[Neenaz Ichaporia] What evidence is there for the efficacy of AI powered learning solutions?
[Mariano Felice] Well, I think that, um, Elena made, um, fantastic points.
[Mariano Felice] I tend to see this question from from 2 different perspectives, to be honest.
[Mariano Felice] I see it, uh, from a technical perspective first and then from a more educational perspective, if you like. So I think that, for example, from from the more technical point of view,
[Mariano Felice] researchers and and developers of AA tools,
[Mariano Felice] as part of the development of of a tool, they have to carry out evaluation to make sure that these tools perform as as they expect, as we expect.
[Mariano Felice] So for example, this means, let's say, in the context of automated scoring, you have to make sure that this system
[Mariano Felice] can technically achieve accuracy that is as close as possible to human accuracy so that that you can use this tool
[Mariano Felice] as as an alternative to to human accuracy. So there's that,
[Mariano Felice] what I would call the the technical
[Mariano Felice] point of view, and there's a lot of evidence for that. Because as I said,
[Mariano Felice] researchers and developers, they have to to do this as part of development. And in the case of researchers, they publish papers where you can see all these evidence. They can see how well these things work. So that's a technical point of view. But then I feel that from the educational point of view, which is more in line with what Elena said, I
[Mariano Felice] think we still have to to to prove,
[Mariano Felice] um,
[Mariano Felice] many of the things that have been mentioned.
[Mariano Felice] Um, I think
[Mariano Felice] I personally feel we need more
[Mariano Felice] evidence of how these tools are effective
[Mariano Felice] at
[Mariano Felice] helping learners achieve their learning goals. And at the moment, I feel that there are very few
[Mariano Felice] studies, uh, in this regard and most of them are qualitative and maybe based on perceptions
[Mariano Felice] or or interviews and so on, which is fine,
[Mariano Felice] but I personally would like to see more objective,
[Mariano Felice] quantitative,
[Mariano Felice] maybe longitudinal
[Mariano Felice] studies
[Mariano Felice] looking at the long term effects and implications.
[Mariano Felice] And I think that I mean, that's the reason why we don't have this evidence now is because we need we need time. Right? We need to see the effect of these tools, which are very new, and we're still exploring ways
[Mariano Felice] to creatively
[Mariano Felice] use them,
[Mariano Felice] um, in order to see how how these tools,
[Mariano Felice] um, have an effect on learning goals and so on. So that is the kind of evidence that I feel we're still lacking and, hopefully, we'll see more of in the future.
[Neenaz Ichaporia] That's absolutely fascinating, Mariano. And, again, aligns
[Neenaz Ichaporia] very much with the research that the British Council carried out, which shows that
[Neenaz Ichaporia] we need more evidence of whether the long term effects actually, there are benefits that are longer term or not. So thank you for that very timely reminder. And and on that note, um, we've spent quite a bit of time considering the benefits of AI use in language learning and teaching, but no debate on AI and education is complete without also considering the risks. So
[Neenaz Ichaporia] I'd like to turn to Guy now.
[Neenaz Ichaporia] Guy, can you tell us what are some of the risks of using AI in language education,
[Neenaz Ichaporia] and what should learners and teachers be mindful of?
[Guy Zaslavsky] So I think, uh,
[Guy Zaslavsky] language learning,
[Guy Zaslavsky] uh, is
[Guy Zaslavsky] a sphere where there are numerous risks
[Guy Zaslavsky] that, uh, can arise by incorporating AI. There are also so many different ways in which AI can get
[Guy Zaslavsky] incorporated.
[Guy Zaslavsky] Uh, some of the top ones that would, uh, top ones that would come to mind,
[Guy Zaslavsky] uh,
[Guy Zaslavsky] let's say, from the learner's perspective,
[Guy Zaslavsky] there is the obvious risk of kind of shooting themselves in the foot. So overusing
[Guy Zaslavsky] all kind of AI aids
[Guy Zaslavsky] to cut corners, cheat in their basically sabotage their own learning progress.
[Guy Zaslavsky] And then,
[Guy Zaslavsky] um, this could be mediated by systems that are are less
[Guy Zaslavsky] prone to such
[Guy Zaslavsky] gaming. So, basically,
[Guy Zaslavsky] adapting the system, you know, obviously, writing essays that used to be sort of prone, sort of resistant
[Guy Zaslavsky] to, uh, to such cheating or with the advent of LLMs, it's become completely broken.
[Guy Zaslavsky] Um, and then,
[Guy Zaslavsky] um, as homework.
[Guy Zaslavsky] Then there is the data privacy risk, which is kind of more of generic thing, but I would say in the context here of children speaking and maybe even having video conversations with systems that are in the cloud.
[Guy Zaslavsky] If they're operated by some
[Guy Zaslavsky] player, you can imagine all sorts of nightmare scenarios where where, uh, an AI system could really
[Guy Zaslavsky] readily understand, for example, what's the sexual orientation
[Guy Zaslavsky] of the child and maybe extort that child if they're still in the closet. So, as you can imagine, so many,
[Guy Zaslavsky] uh, bad things happening, uh, with the through the data privacy,
[Guy Zaslavsky] um, problem that some that AI systems that create.
[Guy Zaslavsky] Uh, from the teacher's perspective, I would say,
[Guy Zaslavsky] there are, of course, the the problems of bias. So,
[Guy Zaslavsky] um, if they're using a system, let's take as an example,
[Guy Zaslavsky] uh, teaching using imagery,
[Guy Zaslavsky] uh, for teaching nouns, so you're teaching about
[Guy Zaslavsky] what is a nurse. And then more most
[Guy Zaslavsky] likely, if you use generative AI to create an image of a nurse, it'll be a female nurse. So this kind of perpetuates this kind of gender bias in the in the culture that nurses are female.
[Guy Zaslavsky] And there are numerous types of biases,
[Guy Zaslavsky] um,
[Guy Zaslavsky] that affect various AI systems. There are also risks of developing overdependence
[Guy Zaslavsky] on AI,
[Guy Zaslavsky] and I would say more, um, a kind of risk of the shifting in authority
[Guy Zaslavsky] so that AI systems will become more competent as teachers than the flesh and blood teacher.
[Guy Zaslavsky] And then it brings all sorts of
[Guy Zaslavsky] questions and problems and risks into how how do you handle such a situation.
[Guy Zaslavsky] Um, that's it.
[Neenaz Ichaporia] Yeah. No. Thank you so much, Guy. That was a very, very, um,
[Neenaz Ichaporia] well summarized
[Neenaz Ichaporia] kind of list of some of the key risks, as you said, particularly around things like data privacy, which I know that, you know, concerns
[Neenaz Ichaporia] many of us and which which is foremost in the minds of many people in the webinar today.
[Neenaz Ichaporia] Um, that's a very timely reminder of some of the risks of AI.
[Neenaz Ichaporia] And, Guy, you touched upon the idea of, uh, the shifting role of the teacher and and moving away from sort of the teacher central to education. And and there's a risk around that. So I'd like to, on that note, come to Niti. Niti, I'd like to ask you, um, this question.
[Neenaz Ichaporia] What can teachers offer that AI can't? What are the limitations of AI?
[Neeti Tripathi] Well, thank you so much for asking that question. I really love that question.
[Neeti Tripathi] Uh, so I want to start with telling, uh, one of the things. We were conducting a research in our school, and we, uh, conducted a survey. And, uh, one of the students wrote that if all the schools start using AI, the teachers are going to lose their job.
[Neeti Tripathi] So, you know, when I was really thinking about it, uh, and I even discussed this with my team, like, why we are not going to lose our jobs because there are many things which we do and AI cannot do. So for example, first of all, the emotional support.
[Neeti Tripathi] A teacher is not only a teacher. A teacher has to play many roles. Sometime we are parent. We are friend. We are adviser.
[Neeti Tripathi] We are we we are mentors.
[Neeti Tripathi] We are guide. We play many role. Uh, you know, we provide that emotional support to our student, which definitely AI cannot do.
[Neeti Tripathi] Then
[Neeti Tripathi] a teacher helps the child to build that relationship not only, uh, you know, with their peers, with the teachers, but also with the entire school community.
[Neeti Tripathi] You know? And the teachers in that process also build relationship with the child and child's family that definitely AI cannot do. This is only done by the teachers.
[Neeti Tripathi] I also feel teachers are very adaptive in their nature. You know? In the class, if they suddenly feel this lesson is not going and I have to change something or my students are not understanding this, this methodology is not working with them, they can come up with something instantly which can go well with the class. I feel this can only be done by the teachers and not
[Neeti Tripathi] AI. Then okay. Again, I would talk about the moral guidance, which, again, you know, teachers always provide and they give and they they they they portray that kind of image. It's a very,
[Neeti Tripathi] um, decent and honest and, you know, uh, ethically, morally acknowledged
[Neeti Tripathi] profession, which
[Neeti Tripathi] only, uh, we educators know and feel about it. So that kind of thing again comes only from us human teachers, not from AI.
[Neeti Tripathi] I feel also the teachers are great role models. You know? Um, they can encourage that love in among their students that see, you know, uh, I'm I can do it, so you can also do it. There are time when the students just give up. They don't want to do things. Again, teachers come helping them as a support, being a role model to them, offering all sort of help and support.
[Neeti Tripathi] Also, if there are conflicts in the class, students are fighting, they're having issues among themselves, who's gonna solve it? Definitely the teachers. So, again, this is one thing that definitely we are going to do and not the AI.
[Neeti Tripathi] And I also feel, you know, uh, I'm talking from a very practical scenario that sometimes there are time when students,
[Neeti Tripathi] they just lose interest. We see many students in the class
[Neeti Tripathi] not wanting to learn, not wanting to study, not wanting to participate. So that time, the teacher comes very handy in creating the curiosity,
[Neeti Tripathi] sparking a love for learning,
[Neeti Tripathi] motivating them, engaging them.
[Neeti Tripathi] So, you know, uh, these things are done only by the teachers. And one of the students once told me that I use chat GPT a lot. I'll I ask a lot of questions because it doesn't judge me. So, yes, I agree that JPG doesn't judge you and can ask a lot of question, but I still feel it's a very flat and
[Neeti Tripathi] very,
[Neeti Tripathi] uh, stable relationship. And but with when it comes to the relationship between the teachers and the students, there are many emotions
[Neeti Tripathi] involved.
[Neeti Tripathi] Those humanly
[Neeti Tripathi] emotions
[Neeti Tripathi] which sometimes take a relationship to a different direction, that that is not done by AI. So definitely, I don't see AI replacing the teachers, and I don't see us losing our job because of the AI.
[Neenaz Ichaporia] Thank you so much, Niti, for that really passionate defense of why teachers are still central and really important to education.
[Neenaz Ichaporia] Um, I am just conscious of time. So I want to I want to turn turn to Mariano with, um, an audience question, actually.
[Neenaz Ichaporia] Chandula in the chat asks about AI and the validity and reliability of testing and evaluation. This is something which Mariano is someone with deep expertise in the applications of AI to language assessment.
[Neenaz Ichaporia] I'd like to ask you. Obviously,
[Neenaz Ichaporia] clearly, accuracy and reliability are very important. So what about bias and inaccuracy in AI? Is this something that language teaching professionals should be alert to?
[Mariano Felice] Yes. Absolutely.
[Mariano Felice] So I think we have to to remind ourselves that these tools are are not perfect, and they're just
[Mariano Felice] as as good as the as the training data
[Mariano Felice] that has been used when they were developed.
[Mariano Felice] So for that reason, we have to be alert to anything that could that could go wrong and especially in an educational context.
[Mariano Felice] So one of the things that we have to bear in mind, for example, is that we should be using tools that have been, uh, trained
[Mariano Felice] for for, um, the same population
[Mariano Felice] that we want to use this for and that means, um, uh,
[Mariano Felice] things such as, uh, the same or similar
[Mariano Felice] culture background, age, level of proficiency, and so on. So we have to make sure these tools have been have been, um, targeted at the same pop population that we're going to use this with. And
[Mariano Felice] to to make sure that this is the case, for example, we can look at,
[Mariano Felice] uh, things such as model cards. In the case of some tools, they provide documentation,
[Mariano Felice] uh, where where you can learn more about how the tools were developed.
[Mariano Felice] Also, I think it's crucial that,
[Mariano Felice] um, teachers and educators
[Mariano Felice] test the tools, uh, they want to use in advance.
[Mariano Felice] So,
[Mariano Felice] um, avoid using the tools for the first time or experimenting
[Mariano Felice] in in the classroom setting. Try that out first. Make sure that they work as expected.
[Mariano Felice] Uh, then as you use the tools,
[Mariano Felice] uh, monitor their use.
[Mariano Felice] If you recommend
[Mariano Felice] a tool, for example, to to your students,
[Mariano Felice] make sure that that they share their experiences with you,
[Mariano Felice] that you even have a look at the output that they're getting from from the AI systems,
[Mariano Felice] that it, uh, aligns with your expectations.
[Mariano Felice] That's the that's one of the ways in which you can, uh, raise any issues or see if there are any red flags. Let's say, for example, if if, um, a tool makes an assessment of of the level of ability of one of your students
[Mariano Felice] and, you know, that that doesn't really align with the kind of proficiency that you see they have in your lessons and so on. So that could be a red flag. So there might be something wrong with the tool or they might not be using it properly and so on. So the more that you
[Mariano Felice] that you, um, use the tools yourselves, you test them, you monitor
[Mariano Felice] their use, uh, the more you will be able to to see if there is anything wrong with them and take action, which I think is crucial.
[Mariano Felice] So you might decide to maybe change the tools or change the task or or make some tweaks, uh, here and there. And for that reason, I think that, um, AI literacy is is key.
[Mariano Felice] So the more that we can learn about how these systems work and so on, the more we are aware
[Mariano Felice] of the risks and what we can do about them.
[Neenaz Ichaporia] Absolutely, Mariano. And thank you so much for giving us a sort of a reckoner of things to keep in mind when we're trying out a new AI tool and how to kind of judge its
[Neenaz Ichaporia] suitability for our particular context.
[Neenaz Ichaporia] There are lots and lots of comments in the chat. More many more than I can keep up with. Or and thank you so much, everyone, for your contributions. I think as one of my colleagues said, if we don't get to every question, we will follow-up with an email later with more resources.
[Neenaz Ichaporia] So look out for that email when it comes to your inbox.
[Neenaz Ichaporia] Um, I would like to very, very quickly, Eleanor, before we go to closing statements,
[Neenaz Ichaporia] if you could come in here, um, just on this point about ethics and AI,
[Neenaz Ichaporia] do teachers and donors need to worry about the ethics of AI? How relevant is it to them? Yeah. Absolutely.
[Elena Oncevska Ager] We should all be concerned about how ethical our engagements are in society more generally and,
[Elena Oncevska Ager] our engagements with AI are no exception.
[Elena Oncevska Ager] Um yes I am worried about students passing off AI generated essays as their own because it means that they have missed out on opportunities to actively engage in developing their ideas and enjoying the process and experiencing
[Elena Oncevska Ager] that joy of learning, that satisfaction of a of a job well done, which I think is central not only to learning but also to well-being.
[Elena Oncevska Ager] Equally though, I'd be worried about teachers being okay with 80% of their lesson planning being done for them. This is, uh, this was echoed in one of the comments in the chat, especially as because I consider lesson planning to be central to teaching vocation. So carefully designing engaging learning tasks for a specific unique
[Elena Oncevska Ager] your own group of students is a privilege and an opportunity to demonstrate that you care that you are there for them to support them in their learning and beyond.
[Elena Oncevska Ager] And I suspect teachers openly talk to their students about lesson planning and their lesson planning co authorship, shall I say, with the AI. Perhaps this is due to the morality of it.
[Elena Oncevska Ager] So integrity is the key ethical consideration.
[Elena Oncevska Ager] Um, other considerations that I will I'm just gonna gloss over in in the interest of time. Uh, we talked about privacy. We talked about, um,
[Elena Oncevska Ager] uh, whether everyone has equal access
[Elena Oncevska Ager] to these technologies,
[Elena Oncevska Ager] whether some cultures are linguistic or
[Elena Oncevska Ager] otherwise are
[Elena Oncevska Ager] underrepresented
[Elena Oncevska Ager] or favored.
[Elena Oncevska Ager] Um, is AI generated
[Elena Oncevska Ager] generated content used responsibly?
[Elena Oncevska Ager] That's another thing that I think we should be thinking about.
[Elena Oncevska Ager] Are people engaging with AI
[Elena Oncevska Ager] in, uh, such a way as to so as to enhance a meaningful human connection?
[Elena Oncevska Ager] Um, are we treating people who engage with AI compassionately and with respect? That is this whole person and not just numbers. Um, the environmental impact, lots of ethical considerations with with just a touch of some. Thank you, Elena. Yeah. Absolutely. And that's that could be another entire session on its own. Absolutely. Well, thank you for the summary.
[Neenaz Ichaporia] Um, now in this final part of the debate, each panelist will have 1 minute to state their final opinion on our question.
[Neenaz Ichaporia] Is AI ultimately a force for good in language education?
[Neenaz Ichaporia] Our panelists, I'm going to have to interrupt you in the interest of time if you go over a minute here. So let's go to Guy first.
[Neenaz Ichaporia] Um, Guy, what are your parting thoughts for the audience on this topic? Is AI ultimately a force for good?
[Guy Zaslavsky] Thank you, James.
[Guy Zaslavsky] Yeah. I think it is. Uh, of course, it it's going to be offering increasingly greater power. And with that, uh, we'll should, hopefully,
[Guy Zaslavsky] this, uh, should be accompanied by great responsibility.
[Guy Zaslavsky] I also think that there is the,
[Guy Zaslavsky] uh, possibility that in the not too distant future, AI will sort of pull the carpet from underneath the line language learning market by offering things such as universal translation. And just like Niti mentioned before,
[Guy Zaslavsky] there would be a growing problem of loss of motivation kids will think
[Guy Zaslavsky] or adults. Why do why should I learn this language? Like, you get Google, like, real time Google translate overlaid
[Guy Zaslavsky] through your AR glasses
[Guy Zaslavsky] for the least majority of the population.
[Guy Zaslavsky] So I think we're going to see just gigantic transformations,
[Guy Zaslavsky] um, facilitated
[Guy Zaslavsky] by, uh, AI.
[Neenaz Ichaporia] Thank you, Guy. That's a bold prediction, and,
[Neenaz Ichaporia] um, you know, it would be very interesting to see how this plays out, um, in the coming months years.
[Neenaz Ichaporia] Um, let me go to Mariano then. Mariano,
[Neenaz Ichaporia] um, what are your, um, what's your take on AI and language education?
[Mariano Felice] Well, I think we have to remember that technology
[Mariano Felice] is what we make of it. It's it's not necessarily good or bad in itself, so it's what we do with it.
[Mariano Felice] And I think that because it's all very new and we're still exploring,
[Mariano Felice] we're a bit uncertain
[Mariano Felice] of of how this will play play out. But
[Mariano Felice] I think it's up to us, and and I think today we've we've touched on
[Mariano Felice] amazing ways of using AI positively.
[Mariano Felice] And I think that as long as we as we keep doing that and we are alert to things that could go wrong and take action to to minimize those risk, then we are we are on
[Mariano Felice] on good track.
[Neenaz Ichaporia] Thank you, Mariano.
[Neenaz Ichaporia] Um, I think that's
[Neenaz Ichaporia] that's a very eloquent reminder, and,
[Neenaz Ichaporia] uh, in the comments has said something similar.
[Neenaz Ichaporia] AI is definitely just a tool, and it's up to us how we use it. But echoing what you just said, Mariano, as well.
[Neenaz Ichaporia] Neeti, please share your takeaways from today's debate.
[Neeti Tripathi] Yes. So for me, yes. AI is there. It's wonderful,
[Neeti Tripathi] amazing, taking care of all the learners, providing personalized
[Neeti Tripathi] feedback,
[Neeti Tripathi] creating differently at the instruction,
[Neeti Tripathi] and teachers.
[Neeti Tripathi] But for me,
[Neeti Tripathi] I have to see, uh, one bottom line, and that's it, pedagogy before technology.
[Neeti Tripathi] AI is here to stay, but it is here to assist us, not to replace it.
[Neeti Tripathi] That's for me. So I would always encourage pedagogy before technology in the field of education.
[Neenaz Ichaporia] Thank you so much, Niti. Um, and and, again, a fabulous reminder, um,
[Neenaz Ichaporia] and echoing what we've been talking about throughout, um, which is a sent centrality and the importance of of human teachers as well. Um, Elena,
[Neenaz Ichaporia] what thoughts would you like to leave the audience with today?
[Elena Oncevska Ager] Yes. Thank you. Um, just to address the main question, is AI a force for good? Um, my answer would be absolutely it can be
[Elena Oncevska Ager] provided that we use it ethically
[Elena Oncevska Ager] in nongenerative,
[Elena Oncevska Ager] nonreplacing
[Elena Oncevska Ager] ways with humans and AI
[Elena Oncevska Ager] playing to each other's strengths like true collab collaborators would,
[Elena Oncevska Ager] um, to jointly support learn learner centered pedagogies
[Elena Oncevska Ager] to help develop agency
[Elena Oncevska Ager] and to create safe nonjudgmental
[Elena Oncevska Ager] spaces for meaning making, hopefully, again,
[Elena Oncevska Ager] working together throughout
[Elena Oncevska Ager] hybrid intelligence
[Elena Oncevska Ager] style.
[Elena Oncevska Ager] Um, also,
[Elena Oncevska Ager] AI can be a force for good provided people are empowered to see
[Elena Oncevska Ager] through the fog, I would say,
[Elena Oncevska Ager] of shiny attractive AI offerings, uh, all the shiny and attractive AI offerings, uh, that promise but don't always deliver on on these ideals.
[Elena Oncevska Ager] Only then I think we can emerge from our collaborations
[Elena Oncevska Ager] with AI feeling
[Elena Oncevska Ager] better prepared and
[Elena Oncevska Ager] human.
[Neenaz Ichaporia] Thank you so much, Elena.
[Neenaz Ichaporia] Um, that point you made about collaboration,
[Neenaz Ichaporia] um, and playing to human and AI strengths is really, really key one.
[Neenaz Ichaporia] Thank you so much to our panelists on behalf of the British Council and the Eltons for sharing your expertise with us and for leaving us with such insightful and really thought provoking ideas.
[Neenaz Ichaporia] But audience, we're not quite done yet. It's now time for you once again to have your say. So our next poll should appear on your screen in a moment, and you'll be able to see the results displayed as well in real time.
[Neenaz Ichaporia] So our next poll is coming up on your screen
[Neenaz Ichaporia] now.
[Neenaz Ichaporia] And this poll is for the teachers in the room.
[Neenaz Ichaporia] And the question being asked is,
[Neenaz Ichaporia] will AI change your role for the better?
[Neenaz Ichaporia] Okay. So I think, um, we can share the poll results and see what our audience is saying.
[Neenaz Ichaporia] Wow.
[Neenaz Ichaporia] Um, there seems to be an overwhelming
[Neenaz Ichaporia] agreement,
[Neenaz Ichaporia] um, certainly amongst the educators that AI will change their role for the better. So a lot of optimism in this room,
[Neenaz Ichaporia] um, and quite a few people are saying maybe or I'm still not sure which shows the need for us to debate issues like this so that you can take an informed opinion.
[Neenaz Ichaporia] Um,
[Neenaz Ichaporia] right. There's one more poll, and this is for the learners.
[Neenaz Ichaporia] Um, and teachers, you can also try to answer this one from the perspective of your learners.
[Neenaz Ichaporia] Um, so that's poll is coming up on your screen now. And the question is, will AI make your language learning journey easier?
[Neenaz Ichaporia] So once again, I think we can publish the full results.
[Neenaz Ichaporia] Right. And even more overwhelming majority with this post, 75%
[Neenaz Ichaporia] of you saying, yes, AI will make your language learning journey easier.
[Neenaz Ichaporia] Again, a few of you saying maybe I'm still not sure, but certainly in this question, um, the majority seems to have decided that AI will make language learning easier for learners.
[Neenaz Ichaporia] Um, just to summarize, we've we've shared so many key ideas today. Um, I think some of the key takeaways for me from this session
[Neenaz Ichaporia] were that AI can provide a vast range of personalized safe environments if used ethically and if used, um, in a way that is suitable for your context.
[Neenaz Ichaporia] Um, the words that jumped out at me were that AI can,
[Neenaz Ichaporia] under the right conditions, be used to create memorable and engaging
[Neenaz Ichaporia] experiences for learners.
[Neenaz Ichaporia] The promise of differentiated learning, of course, there are issues around data privacy. There are issues around, um, overuse or overreliance on AI. There is more evidence needed of the efficacy of AI longer term.
[Neenaz Ichaporia] But I think the words I'd like to maybe leave our audience with today are yours, Niti, which is pedagogy before technology.
[Neenaz Ichaporia] Um, and I think that that was a very timely reminder for all of us,
[Neenaz Ichaporia] um, in today's debate.
[Neenaz Ichaporia] Once again, thank you so much to the audience
[Neenaz Ichaporia] and our speakers
[Neenaz Ichaporia] for joining us and giving us our time, um, today.
[Neenaz Ichaporia] In a moment, you'll get a pop up survey on your screen. Please take a couple of minutes to share your thoughts. It really helps us plan and deliver even better sessions in the future.
[Neenaz Ichaporia] And up next,
[Neenaz Ichaporia] please head back to the festival platform for sessions on embracing inclusive practices
[Neenaz Ichaporia] and what you didn't know about online learning. Those sessions are starting
[Neenaz Ichaporia] very shortly.
[Neenaz Ichaporia] Um, so do head back to the festival platform
[Neenaz Ichaporia] for those sessions as well. Thank you very much for joining us today.