WEBVTT 00:00:08.300 --> 00:00:11.820 Opening AI for Language Learning explores artificial 00:00:11.820 --> 00:00:14.480 intelligence through informal conversations between 00:00:14.480 --> 00:00:17.019 two longtime colleagues in language education 00:00:17.019 --> 00:00:20.940 and digital technologies, Mat Schulze and Phil 00:00:20.940 --> 00:00:25.359 Hubbard. Language educators, learners, and enthusiasts 00:00:25.359 --> 00:00:28.160 are invited to an informed discussion where emerging 00:00:28.160 --> 00:00:33.840 technology meets human communication. All right. 00:00:33.920 --> 00:00:37.840 Welcome back to Opening AI for Language Learning. 00:00:38.060 --> 00:00:42.700 I'm Phil. And my name is Mat. Over the last 00:00:42.700 --> 00:00:46.679 three episodes, we talked particularly about 00:00:46.679 --> 00:00:52.219 GenAI stuff you need to know now. We introduced 00:00:52.219 --> 00:00:55.020 the large language models and we kind of based 00:00:55.020 --> 00:01:01.340 all this on a paper both Phil and I wrote. So 00:01:01.340 --> 00:01:04.060 you can catch up with those things if you haven't 00:01:04.060 --> 00:01:07.510 heard them. They're available as our podcast, 00:01:07.530 --> 00:01:10.370 and of course you can also read the paper we 00:01:10.370 --> 00:01:14.310 are basing our conversations on. The links will 00:01:14.310 --> 00:01:17.849 be in the notes and they'll be easy to find. 00:01:17.849 --> 00:01:19.890 It's a little bit difficult to do this orally. 00:01:20.689 --> 00:01:25.409 What are we going to do today, Phil? Well, we've 00:01:25.409 --> 00:01:28.370 talked about what goes into GenAI competency 00:01:28.370 --> 00:01:31.209 for language teachers. Today we're going to talk 00:01:31.209 --> 00:01:34.049 about how do you get there? How do you develop 00:01:34.049 --> 00:01:37.769 it? So first, we're going to motivate the need 00:01:37.769 --> 00:01:41.549 for ongoing professional development. The idea 00:01:41.549 --> 00:01:48.129 that this is really never ending. We'll describe 00:01:48.129 --> 00:02:00.030 our framework and this idea of GEN -AI SIPD, 00:02:00.250 --> 00:02:03.209 so sustained integrated professional development. 00:02:04.980 --> 00:02:08.080 Then for new work here, we're going to go over 00:02:08.080 --> 00:02:11.819 seven provisional principles for GEN -AI SIPD. 00:02:12.259 --> 00:02:14.120 And then we'll talk a little bit about how we're 00:02:14.120 --> 00:02:20.419 doing GEN -AI SIPD for ourselves. Yeah, this 00:02:20.419 --> 00:02:23.919 is this kind of doing it ourselves. This is how 00:02:23.919 --> 00:02:27.159 this all started. This is how we started our 00:02:28.069 --> 00:02:30.169 conversations. We had finished writing the paper. 00:02:30.210 --> 00:02:32.490 Of course, we kind of met occasionally, or actually 00:02:32.490 --> 00:02:36.449 quite frequently, when writing the paper and 00:02:36.449 --> 00:02:40.849 we decided both of us to continue with these 00:02:40.849 --> 00:02:43.189 meetings. And of course, the topic always was 00:02:43.189 --> 00:02:46.129 Gen. AI, as it is the topic, I believe. I was 00:02:46.129 --> 00:02:49.830 talking to a larger group of teachers yesterday. 00:02:50.009 --> 00:02:53.110 It's for many teachers and for many people, it's 00:02:53.110 --> 00:02:57.400 a kind of big topic these days. We kept talking 00:02:57.400 --> 00:02:59.680 about it, we kept learning about it, and at some 00:02:59.680 --> 00:03:01.479 stage we basically thought, why don't we share 00:03:01.479 --> 00:03:04.919 this? And you're listening to the outcome of 00:03:04.919 --> 00:03:08.099 that kind of vague idea. All right, we're moving 00:03:08.099 --> 00:03:11.759 on to those principles. Yeah, we thought we'd 00:03:11.759 --> 00:03:13.740 go ahead and lead off with the principles and 00:03:13.740 --> 00:03:17.060 then tell you a little bit afterwards how we 00:03:17.060 --> 00:03:21.580 got there. So the first principle is built from 00:03:21.580 --> 00:03:24.639 the first part of the paper. And that's to understand 00:03:24.639 --> 00:03:27.500 the basics of GenAI so that you can make these 00:03:27.500 --> 00:03:31.639 in use decisions in an informed way. And that 00:03:31.639 --> 00:03:35.840 includes both understanding what large language 00:03:35.840 --> 00:03:38.400 models are, a little bit about where they came 00:03:38.400 --> 00:03:43.159 from, how they work and such, and then also the 00:03:43.159 --> 00:03:49.740 idea that there are these 10 areas that we think 00:03:49.740 --> 00:03:52.729 you should get some knowledge in early on. The 00:03:52.729 --> 00:03:56.050 second one relates to learner training, and it's 00:03:56.050 --> 00:03:59.550 the idea of experiencing gen AI from the learner's 00:03:59.550 --> 00:04:03.449 perspective. Being a language learner yourself 00:04:03.449 --> 00:04:07.409 with AI, going through the material in AI that 00:04:07.409 --> 00:04:11.930 you might produce for the learners as if you 00:04:11.930 --> 00:04:14.889 were a learner, even though the AI produced it 00:04:14.889 --> 00:04:18.610 in the target language, and then talking with 00:04:18.610 --> 00:04:22.259 the learners about what's going on so that you 00:04:22.259 --> 00:04:26.480 get their perspective. The third one, start small. 00:04:26.819 --> 00:04:29.600 Look for uses of GEN -AI that have some immediate 00:04:29.600 --> 00:04:33.620 value. This is the integrated part of the sustained 00:04:33.620 --> 00:04:37.819 integrated. But rather than go off on a larger 00:04:37.819 --> 00:04:41.879 project or work with materials from webinars 00:04:41.879 --> 00:04:44.519 and so on that you can't see a clear connection 00:04:44.519 --> 00:04:48.259 with your own work, find the low -hanging fruit. 00:04:48.439 --> 00:04:51.939 and get into AI for language teaching through 00:04:51.939 --> 00:04:58.120 that. The fourth one is thinking of this SIPD 00:04:58.120 --> 00:05:04.779 as being pronounced SIPD. So GenAI SIPD for language 00:05:04.779 --> 00:05:10.439 teaching means taking little bits of it, consistent, 00:05:10.660 --> 00:05:13.959 maybe even daily micro learning, in order to 00:05:13.959 --> 00:05:20.480 get this a habit, a change of the way you think 00:05:20.480 --> 00:05:27.240 about AI as not just an add -on. The fifth, very 00:05:27.240 --> 00:05:30.800 important one, to evaluate your AI options critically 00:05:30.800 --> 00:05:33.860 and reflectively. Now, we all should be doing 00:05:33.860 --> 00:05:38.620 that for everything new that we bring into language 00:05:38.620 --> 00:05:42.720 teaching, but because AI is so new, it's often 00:05:42.720 --> 00:05:45.720 very hard to do that unless there's a conscious 00:05:45.720 --> 00:05:49.939 effort made. It can't be done in a more automatic 00:05:49.939 --> 00:05:52.060 way the way we would if we looked at a textbook 00:05:52.060 --> 00:05:54.459 and we could see immediately, oh, here are the 00:05:54.459 --> 00:05:56.939 parts I want to use and how I want to use it. 00:05:57.720 --> 00:06:00.720 The sixth one, put ethical issues front and center. 00:06:00.879 --> 00:06:04.480 Always consider those in your students' GEN -AI 00:06:04.480 --> 00:06:08.980 use and in your own. And then finally, the seventh 00:06:08.980 --> 00:06:12.040 one, seek out and nurture a partnership with 00:06:12.040 --> 00:06:14.740 peers. in much the same way that Mat and I are 00:06:14.740 --> 00:06:18.100 doing it for each other here. Okay, so those 00:06:18.100 --> 00:06:21.680 are the seven principles that we propose to guide 00:06:21.680 --> 00:06:25.420 your GEN .ai journey and that are guiding ours. 00:06:27.040 --> 00:06:31.879 So, yeah, SIAPD is sustained integrated professional 00:06:31.879 --> 00:06:36.600 development. Of course, the question arises why 00:06:36.600 --> 00:06:39.899 are we actually doing this? Why are we proposing 00:06:39.899 --> 00:06:45.139 another idea, concept of professional development, 00:06:45.220 --> 00:06:51.019 particularly with GEN .AI in this case. I personally 00:06:51.019 --> 00:06:53.660 believe that GEN .AI of course is still relatively 00:06:53.660 --> 00:06:55.779 new. There are lots of people who argue we're 00:06:55.779 --> 00:07:00.160 at the very beginning of these radical developments 00:07:00.160 --> 00:07:03.759 in digital technologies and artificial intelligence. 00:07:06.120 --> 00:07:10.800 So it might be that the so -called traditional 00:07:10.810 --> 00:07:14.209 approaches to professional development are no 00:07:14.209 --> 00:07:17.629 longer adequate to gen .ai. Of course, when we 00:07:17.629 --> 00:07:21.069 say traditional, the film has these weird negative 00:07:21.069 --> 00:07:25.050 meanings, certainly in language education, as 00:07:25.050 --> 00:07:28.230 a former colleague of mine pointed out to me 00:07:28.230 --> 00:07:31.050 when she was talking about her latest project, 00:07:31.149 --> 00:07:33.870 where she looks at these traditional approaches 00:07:33.870 --> 00:07:36.829 as they're very often called in applied linguistics 00:07:36.829 --> 00:07:39.069 and language education. It's always meant negative. 00:07:39.310 --> 00:07:42.569 We don't really mean it negative. We're basically 00:07:42.569 --> 00:07:44.990 saying this is a radical new technology that 00:07:44.990 --> 00:07:50.269 is evolving very, very rapidly. So we're double 00:07:50.269 --> 00:07:55.019 checking that. older approach, the approach we're 00:07:55.019 --> 00:07:58.300 all used to in professional development. Is it 00:07:58.300 --> 00:08:00.399 still working or not? Do we need to amend it? 00:08:00.399 --> 00:08:03.259 Do we need to add? Do we need to subtract in 00:08:03.259 --> 00:08:06.480 some way? So the question then arises, what does 00:08:06.480 --> 00:08:08.819 actually make professional development successful? 00:08:08.899 --> 00:08:11.720 How can we do good professional development? 00:08:12.459 --> 00:08:15.740 What do you think, Phil? Well, that's quite the 00:08:15.740 --> 00:08:19.319 question. And luckily, some of our colleagues 00:08:19.319 --> 00:08:24.329 in the past have taken a look at that. As part 00:08:24.329 --> 00:08:29.910 of our journey in producing the paper, we did 00:08:29.910 --> 00:08:33.549 some background reading in professional development 00:08:33.549 --> 00:08:38.509 and two studies in particular stood out. The 00:08:38.509 --> 00:08:42.570 one that I consider the flagship by Linda Darling 00:08:42.570 --> 00:08:46.309 Hammond and colleagues from 2017, obviously not 00:08:46.309 --> 00:08:54.149 targeting AI, but they looked at 35 programs, 00:08:54.549 --> 00:08:56.750 professional development programs that had been 00:08:56.750 --> 00:09:02.009 noted as being successful in producing concrete 00:09:02.009 --> 00:09:07.450 outcomes. And from that, they extracted seven 00:09:07.450 --> 00:09:12.049 components of successful PD programs. For our 00:09:12.049 --> 00:09:14.889 purposes, because ours is a little more oriented 00:09:14.889 --> 00:09:17.889 towards self -directed learning, or at least 00:09:17.889 --> 00:09:22.129 compatible with it, rather than top -down structured 00:09:22.129 --> 00:09:27.649 PD programs. They said integration into the classroom, 00:09:27.929 --> 00:09:30.629 into professional practice, that's our integrated 00:09:30.629 --> 00:09:35.090 part, critical reflection, which we think should 00:09:35.090 --> 00:09:37.629 subsume all of this, but of course is also our 00:09:37.629 --> 00:09:40.690 principle number five, and then sustained effort, 00:09:41.549 --> 00:09:45.360 the sustained part of the SIPD. And we agree 00:09:45.360 --> 00:09:48.879 with that. Interestingly, a previous paper by 00:09:48.879 --> 00:09:54.039 Matherson et al. in 2014 within the TPAC framework 00:09:54.039 --> 00:09:59.259 had also noted those three areas that integration 00:09:59.259 --> 00:10:01.399 into the classroom and professional practice, 00:10:01.919 --> 00:10:05.159 critical reflection, and sustained effort were 00:10:05.159 --> 00:10:09.279 all key to this sort of professional development. 00:10:11.080 --> 00:10:16.620 Makes a lot of sense. always believe the concept 00:10:16.620 --> 00:10:21.639 of teacher reflection is perhaps the single most 00:10:21.639 --> 00:10:26.379 important indicator and factor in teacher development 00:10:26.379 --> 00:10:30.120 and the personal professional development, not 00:10:30.120 --> 00:10:32.419 the kind of courses and things like that. So 00:10:32.419 --> 00:10:37.419 yeah, there's of course we have the word sustained. 00:10:37.820 --> 00:10:40.460 You just use the fill and we've got the word 00:10:40.460 --> 00:10:44.139 integrated in the sustained integrated professional 00:10:44.139 --> 00:10:48.899 development, the SISIPD. So we're adding sustained 00:10:48.899 --> 00:10:52.500 as an adjective to professional development. 00:10:55.240 --> 00:10:58.500 What we mean by it is kind of to coin a new English 00:10:58.500 --> 00:11:02.399 word, we make it upkeepable basically. So we 00:11:02.399 --> 00:11:05.529 also make it sustainable. by that way. You can 00:11:05.529 --> 00:11:08.370 sustain it if you do it in that way. You can 00:11:08.370 --> 00:11:12.929 keep it up, basically. And you should keep it 00:11:12.929 --> 00:11:15.389 up. It means both. That's why we use sustained 00:11:15.389 --> 00:11:21.789 and not sustainable. Integrate it. Do it. This 00:11:21.789 --> 00:11:23.809 gen .ai is on top of everything else. Nothing 00:11:23.809 --> 00:11:29.009 has gone away, basically, when ChatGPT was launched 00:11:29.009 --> 00:11:37.309 in the mainstream area. So do this kind of professional 00:11:37.309 --> 00:11:39.970 development pretty much when you do something 00:11:39.970 --> 00:11:43.809 else that you perhaps have to do or want to do. 00:11:44.750 --> 00:11:48.669 And that's what we practically mean by integrated 00:11:48.669 --> 00:11:52.309 professional development. And what's been important 00:11:52.309 --> 00:11:57.129 for us is that this is a targeted version of 00:11:57.129 --> 00:12:00.690 what's commonly called continuing or better, 00:12:00.889 --> 00:12:03.480 I think more accurately. continuous professional 00:12:03.480 --> 00:12:10.379 development, our GEN -AI SIPD. It's specific 00:12:10.379 --> 00:12:13.500 to language teaching and learning, and that's 00:12:13.500 --> 00:12:16.940 hard to find out there. We found a few examples 00:12:16.940 --> 00:12:23.360 of proposals, though, and a lot of them are very 00:12:23.360 --> 00:12:25.580 similar to what you see for general education. 00:12:26.139 --> 00:12:29.139 And ours kind of starts that way too. We based 00:12:29.139 --> 00:12:34.480 it at some degree on like the UNESCO 2024 framework, 00:12:34.500 --> 00:12:37.600 but we always want to interpret it through this 00:12:37.600 --> 00:12:41.200 language teaching and learning lens. The second 00:12:41.200 --> 00:12:45.879 point is that this is ongoing. That's the sustained 00:12:45.879 --> 00:12:50.159 part, but it's also largely self directed. That 00:12:50.159 --> 00:12:54.360 doesn't mean independent or certainly. completely 00:12:54.360 --> 00:12:56.639 on your own without interacting with others. 00:12:56.919 --> 00:13:01.139 It doesn't mean we don't encourage going to webinars 00:13:01.139 --> 00:13:05.019 and things like that, but it means that as the 00:13:05.019 --> 00:13:09.600 teacher, you know your teaching context and you 00:13:09.600 --> 00:13:12.600 want to make this directly relevant as much as 00:13:12.600 --> 00:13:19.220 possible. So ultimately, this entails a shift 00:13:19.220 --> 00:13:22.620 in your mindset. It certainly has changed my 00:13:22.620 --> 00:13:27.100 mindset. And I suspect Mat's as well, but I'll 00:13:27.100 --> 00:13:32.500 leave that up to him. I change my mindset all 00:13:32.500 --> 00:13:36.019 the time. Well, the main thing is you're not 00:13:36.019 --> 00:13:38.620 completely set in your mindset. And that's the 00:13:38.620 --> 00:13:41.279 idea. We're constantly seeking to understand 00:13:41.279 --> 00:13:47.159 and utilize GenAI, both as a tool and one used 00:13:47.159 --> 00:13:51.399 critically, but not something that we back off 00:13:51.399 --> 00:13:55.679 from trying out. And as a collaborator is what 00:13:55.679 --> 00:14:02.759 Ethan Molek, one of my heroes, I guess, of educational 00:14:02.759 --> 00:14:08.679 gen AI calls co -intelligence. The metaphor of 00:14:08.679 --> 00:14:11.679 collaboration is good. I think we'll have to 00:14:11.679 --> 00:14:14.000 have another discussion perhaps at some time 00:14:14.000 --> 00:14:16.809 in one of our conversations. Because I've also 00:14:16.809 --> 00:14:18.830 been thinking about it a lot. How do you actually 00:14:18.830 --> 00:14:21.169 collaborate with a machine? Because it's a very 00:14:21.169 --> 00:14:24.389 different level when two humans collaborate versus 00:14:24.389 --> 00:14:26.970 a human and a machine collaborate in some way. 00:14:27.169 --> 00:14:29.710 There are quite a number of things to consider. 00:14:30.009 --> 00:14:34.370 But let's stick with the topic of SIPD. And perhaps, 00:14:34.629 --> 00:14:37.389 Phil, we can pull it all together and look at 00:14:37.389 --> 00:14:42.950 the entire framework of SIPD briefly. Well, I'll 00:14:42.950 --> 00:14:45.720 let you do that. What does the framework consist 00:14:45.720 --> 00:14:51.039 of? Well, all right then. We talked about the 00:14:51.039 --> 00:14:54.679 basic knowledge. So yes, you need a base for 00:14:54.679 --> 00:14:57.299 those things. That's why we started off with 00:14:57.299 --> 00:15:02.080 the LLM. And we have those 10 areas. They're 00:15:02.080 --> 00:15:05.700 both skills and knowledge, items, areas of knowledge, 00:15:06.019 --> 00:15:09.120 areas of skills. And now we're drawing on - The 00:15:09.120 --> 00:15:13.820 GENI stuff you need to know now, basically. And 00:15:13.820 --> 00:15:19.320 now, yes. Yes. And so, in this talk, we're drawing 00:15:19.320 --> 00:15:22.200 out those seven principles, and we'll get back 00:15:22.200 --> 00:15:25.679 to those in a moment, that this drives and guides 00:15:25.679 --> 00:15:31.899 the process of building on that foundation. So, 00:15:31.980 --> 00:15:35.179 of course, this is very practical in many ways 00:15:35.179 --> 00:15:41.129 and as many things in life. And in nature, it's 00:15:41.129 --> 00:15:43.389 non -linear. So it's not kind of you go from 00:15:43.389 --> 00:15:46.190 A to B in a straight line. It's going up and 00:15:46.190 --> 00:15:54.950 down in many different ways. OK. And it's something 00:15:54.950 --> 00:15:58.429 that we're proposing for others. Otherwise, we 00:15:58.429 --> 00:16:01.570 just have our talks as we had been, just the 00:16:01.570 --> 00:16:06.149 two of us, over coffee and not bother to write 00:16:06.149 --> 00:16:08.429 the paper, which was a lot of work, but it was 00:16:08.620 --> 00:16:11.600 also really important. So we're proposing this 00:16:11.600 --> 00:16:15.139 for others, but we're also trying to live it 00:16:15.139 --> 00:16:18.720 ourselves, take our own medicine, get down and 00:16:18.720 --> 00:16:23.240 dirty in the in the mud of this, because in doing 00:16:23.240 --> 00:16:26.919 so, we not only we think add to our credibility, 00:16:27.100 --> 00:16:29.960 but we're actually learning some really cool 00:16:29.960 --> 00:16:35.690 stuff that we didn't know before. So the seven 00:16:35.690 --> 00:16:37.690 principles, I think it kind of bears repeating 00:16:37.690 --> 00:16:41.070 them. I'm not very good at kind of learning by 00:16:41.070 --> 00:16:45.250 heart to going through. Sometimes I have trouble 00:16:45.250 --> 00:16:48.990 kind of getting all seven together if I'm kind 00:16:48.990 --> 00:16:52.370 of quizzed on it, very ad hoc. So the first one 00:16:52.370 --> 00:16:55.370 was understand the basics of Gen .ai. That one 00:16:55.370 --> 00:16:59.549 I kind of always remember because it is. In my 00:16:59.549 --> 00:17:02.190 opinion, very essential to have gained this basic 00:17:02.190 --> 00:17:04.890 understanding. Number two is experience GEN .AI 00:17:04.890 --> 00:17:07.529 from the language learner perspective. I know 00:17:07.529 --> 00:17:11.250 this is one, Phil, you are very keen on, and 00:17:11.250 --> 00:17:13.710 we're also gonna talk about a little bit more 00:17:13.710 --> 00:17:18.910 perhaps. Number three, start small and look for 00:17:18.910 --> 00:17:21.970 uses of GEN .AI that have immediate value. This 00:17:21.970 --> 00:17:25.329 is partially. motivational. I always remember 00:17:25.329 --> 00:17:28.589 Graham Davis, one of the early pioneers of computer 00:17:28.589 --> 00:17:30.730 -assisted language learning. He called it the 00:17:30.730 --> 00:17:36.410 KISS principle. Keep it simple, stupid. It's 00:17:36.410 --> 00:17:39.069 literally about sustaining it, basically, to 00:17:39.069 --> 00:17:43.230 keep it small and successful in some way. So 00:17:43.230 --> 00:17:45.369 looking for the low -hanging fruit, as you said 00:17:45.369 --> 00:17:50.029 earlier. Number four, think of Gen .AI. a GenAI 00:17:50.029 --> 00:17:55.049 SIPD as GenAI SIPD. So again, do it in small 00:17:55.049 --> 00:17:57.289 chunks, but do it very regularly. Do it like 00:17:57.289 --> 00:18:01.009 you sip a coffee. Do it as kind of what is technically 00:18:01.009 --> 00:18:07.930 called micro learning. And number five is kind 00:18:07.930 --> 00:18:10.930 of evaluate your AI options critically and reflectively. 00:18:11.150 --> 00:18:13.650 The reflection part I believe is very important. 00:18:13.789 --> 00:18:16.869 This for me personally is the most important 00:18:16.869 --> 00:18:20.809 of the seven. principles because it will decide 00:18:20.809 --> 00:18:24.529 largely about success or otherwise. Number six, 00:18:24.529 --> 00:18:28.190 always consider ethical issues. That is, of course, 00:18:28.470 --> 00:18:30.869 a huge topic. We're probably going to have a 00:18:30.869 --> 00:18:34.369 separate conversation about it. And number seven, 00:18:34.470 --> 00:18:38.849 as we've been saying and as we are doing, seek 00:18:38.849 --> 00:18:42.630 out and nurture partnerships with peers. Work 00:18:42.630 --> 00:18:45.910 together. You're not in this alone in some way. 00:18:46.049 --> 00:18:49.470 And yes, this can also include formal workshops 00:18:49.470 --> 00:18:56.170 and networks in some way. Okay, so why is it 00:18:56.170 --> 00:19:00.210 so important for us to do something now? Can't 00:19:00.210 --> 00:19:03.289 we just kick this down the road until it's well 00:19:03.289 --> 00:19:07.150 established and the right textbook comes out 00:19:07.150 --> 00:19:10.470 and we can just follow that? I think some people 00:19:10.470 --> 00:19:15.029 think this, but Genia is not going to go away. 00:19:15.150 --> 00:19:18.420 I think that much we know already. even in this 00:19:18.420 --> 00:19:24.140 initial phase. And I would argue if there's change 00:19:24.140 --> 00:19:27.319 around and you're subjected to change in some 00:19:27.319 --> 00:19:30.140 way or affected by change, embrace it. It's the 00:19:30.140 --> 00:19:32.339 much better way. Don't reject it. Don't try to 00:19:32.339 --> 00:19:34.880 block it off. It's hardly ever successful. It's 00:19:34.880 --> 00:19:36.559 like kind of keeping a beach ball underwater, 00:19:36.559 --> 00:19:41.940 basically. It will come up at some stage. Yeah, 00:19:41.940 --> 00:19:47.670 and it changes so rapidly. That's why the even 00:19:47.670 --> 00:19:50.849 our own paper, if we were writing it now, there 00:19:50.849 --> 00:19:53.430 would be certainly some different examples and 00:19:53.430 --> 00:19:55.410 some different literature we would bring in, 00:19:55.809 --> 00:20:00.089 and probably different emphases we would make 00:20:00.089 --> 00:20:03.630 on certain points. And that's okay. Maybe we'll 00:20:03.630 --> 00:20:05.430 come back and do a version two of the paper. 00:20:05.549 --> 00:20:08.509 It would be interesting to see that. Well, and 00:20:08.509 --> 00:20:10.910 we're talking about it now, which helps already 00:20:10.910 --> 00:20:13.250 with kind of thinking about these changes, right? 00:20:13.369 --> 00:20:16.269 That's a good point. Yeah, this is our update. 00:20:17.049 --> 00:20:20.809 Those who listen to these podcasts won't need 00:20:20.809 --> 00:20:23.190 to go back and reread the paper. They can just 00:20:23.190 --> 00:20:24.990 read it once and then listen to what we have 00:20:24.990 --> 00:20:29.170 to say about it. I mentioned Ethan Molek just 00:20:29.170 --> 00:20:32.170 a moment ago, and in his book, Co -Intelligence, 00:20:32.549 --> 00:20:36.990 written in 2024, he has this quote that I love 00:20:36.990 --> 00:20:39.829 and that really keeps me going on this path, 00:20:40.250 --> 00:20:44.900 and that is, whichever AI you're using now will 00:20:44.900 --> 00:20:50.380 be the worst AI you will ever use. So areas that 00:20:50.380 --> 00:20:55.779 you find frustrating or counterproductive or 00:20:55.779 --> 00:20:59.619 just somehow not doing what you want it to do, 00:21:01.440 --> 00:21:03.500 tomorrow those may already be there, some of 00:21:03.500 --> 00:21:08.140 the fixes. And the only way you can do is keep 00:21:08.140 --> 00:21:10.880 tracking that. So now what we're going to do 00:21:10.880 --> 00:21:15.200 is Say a little bit about what we've been doing. 00:21:16.119 --> 00:21:18.519 We'll each take one of the principles and talk 00:21:18.519 --> 00:21:22.599 about how we're using it to drive our gen AI 00:21:22.599 --> 00:21:27.750 journey. Alright, let me yeah I. can go first 00:21:27.750 --> 00:21:32.190 there and give you a breathing space, Phil. I 00:21:32.190 --> 00:21:35.710 already said the principle that I'm most interested 00:21:35.710 --> 00:21:38.369 in and find most valuable is principle number 00:21:38.369 --> 00:21:42.390 five, evaluate AI options critically and reflectively. 00:21:43.549 --> 00:21:47.990 I believe for reflection, it is a better reflection 00:21:47.990 --> 00:21:52.500 if one has a lens. Personally, I like a theoretical 00:21:52.500 --> 00:21:54.880 lens. Of course, you can also have an experiential 00:21:54.880 --> 00:21:58.779 lens. You look at things through a lens of a 00:21:58.779 --> 00:22:02.660 particular kind. I like doing this through reading 00:22:02.660 --> 00:22:04.859 in particular, but also listening to podcasts, 00:22:04.920 --> 00:22:08.079 for example. I subscribe to quite a number of 00:22:08.079 --> 00:22:14.299 blogs on AI and generative AI. Read those, sometimes 00:22:14.299 --> 00:22:18.420 lengthy. blog posts. I collect those papers also 00:22:18.420 --> 00:22:20.960 to be able to revisit them and it's a little 00:22:20.960 --> 00:22:24.319 bit track in progress over time, my own and my 00:22:24.319 --> 00:22:29.380 own learning, but as well the progress and development, 00:22:29.880 --> 00:22:36.119 very fast development of generative AI. And then 00:22:36.119 --> 00:22:38.019 of course it's discussions like these. I'm kind 00:22:38.019 --> 00:22:41.980 of a very extrovert person, so I process best 00:22:41.980 --> 00:22:45.380 when I can talk about things, when I have somebody 00:22:45.380 --> 00:22:48.240 to interact with, then I learn better, so I kind 00:22:48.240 --> 00:22:50.799 of seek those out. I do this also with students 00:22:50.799 --> 00:22:53.259 whenever I can and whenever it's appropriate 00:22:53.259 --> 00:22:56.700 in class, because the best way to learn, very 00:22:56.700 --> 00:22:58.819 often, it's an old cliche, right, is to teach 00:22:58.819 --> 00:23:01.119 it, because then you're better prepared, more 00:23:01.119 --> 00:23:05.140 focused, and doing things. I also kind of do 00:23:05.140 --> 00:23:08.680 professional development workshops on this. The 00:23:08.680 --> 00:23:11.000 one thing I want to... Say because it's always 00:23:11.000 --> 00:23:14.019 important the word critical and we said kind 00:23:14.019 --> 00:23:18.440 of evaluate AI options critically is sometimes 00:23:18.440 --> 00:23:21.339 misunderstood as being negative or even rejecting 00:23:21.339 --> 00:23:24.160 things. That's surely not what we mean. Critical 00:23:24.160 --> 00:23:28.220 means is it fit for purpose? Does it serve me? 00:23:28.480 --> 00:23:32.339 Can I adapt it to my context? That's the critical 00:23:32.339 --> 00:23:37.079 approach, basically. And is there sometimes criticism 00:23:37.079 --> 00:23:40.180 and is it worthwhile criticism about any AI hype 00:23:40.180 --> 00:23:43.160 or things like that? Yes, that might work too. 00:23:43.660 --> 00:23:46.420 I think I'll leave it at that. There's lots more 00:23:46.420 --> 00:23:49.400 to say about reflection, but we can pick another 00:23:49.400 --> 00:23:52.420 principle, I think. Yeah, we might want to do 00:23:52.420 --> 00:23:57.740 a whole episode on critical analysis at some 00:23:57.740 --> 00:24:02.579 point. So my example is from principle number 00:24:02.579 --> 00:24:06.400 two. And I've been working in the area of learner 00:24:06.400 --> 00:24:09.940 training, which is this is experience from the 00:24:09.940 --> 00:24:14.420 learner's perspective. And in order to be a credible 00:24:14.420 --> 00:24:17.559 learner trainer, it seems to me you need to be 00:24:17.559 --> 00:24:22.099 a learner. Or at least that's one very good way 00:24:22.099 --> 00:24:27.910 to do it. So I'm interested in listening. That 00:24:27.910 --> 00:24:30.589 was something I taught for many years before 00:24:30.589 --> 00:24:33.410 my retirement, and I continue being interested 00:24:33.410 --> 00:24:35.869 in it, and especially looking to see what AI 00:24:35.869 --> 00:24:40.750 can do to help that. The example I'll talk about 00:24:40.750 --> 00:24:45.750 is from Google Translate practice mode, which 00:24:45.750 --> 00:24:49.690 is in beta version on my phone. When I open the 00:24:49.690 --> 00:24:52.990 Google Translate app, it's at the bottom right 00:24:52.990 --> 00:24:57.670 -hand corner. It does say beta and warns you 00:24:57.670 --> 00:25:00.789 that AIs can make mistakes. That's the fine print 00:25:00.789 --> 00:25:04.809 that Mat was talking about before. It has a 00:25:04.809 --> 00:25:07.190 very simple user interface. I've been working 00:25:07.190 --> 00:25:10.269 in both French and Spanish for different reasons. 00:25:10.329 --> 00:25:13.170 The French because I have two granddaughters 00:25:13.170 --> 00:25:16.529 who live in French speaking Switzerland and I 00:25:16.529 --> 00:25:19.890 want to be able to experience the culture there 00:25:19.890 --> 00:25:22.390 a little bit better than I have been in the past. 00:25:24.949 --> 00:25:28.470 The user interface for it's very simple. You 00:25:28.470 --> 00:25:33.890 open up a lesson that it has prepared for you. 00:25:33.890 --> 00:25:36.710 It gives you an option of three of them in any 00:25:36.710 --> 00:25:40.309 given day, typically. And ideally you go through 00:25:40.309 --> 00:25:43.789 all three on a topic, a communicative topic that 00:25:43.789 --> 00:25:47.509 you previously said is of value to you from a 00:25:47.509 --> 00:25:51.349 list that they provide. There is a vocabulary 00:25:51.349 --> 00:25:58.130 preview. and then a presentation of a short conversation 00:25:58.130 --> 00:26:02.410 between two interlocutors. There's a speaking 00:26:02.410 --> 00:26:05.029 mode as well. I focus on the listening mode right 00:26:05.029 --> 00:26:07.329 now, but I'll be getting into the speaking part 00:26:07.329 --> 00:26:09.150 eventually because I think it's probably the 00:26:09.150 --> 00:26:13.690 most powerful part. In the listening mode, there 00:26:13.690 --> 00:26:18.269 are two types of items. One of them consists 00:26:18.269 --> 00:26:22.309 of a collection of words. When you listen to 00:26:22.309 --> 00:26:26.049 the conversational turn, two to four sentences 00:26:26.049 --> 00:26:29.690 typically from one speaker, then it says which 00:26:29.690 --> 00:26:31.930 of the words that are listed there did you hear? 00:26:32.029 --> 00:26:35.029 It's usually two or three of them out of as many 00:26:35.029 --> 00:26:40.809 as six or seven. It also has an item type where 00:26:40.809 --> 00:26:44.069 you have a short multiple choice comprehension 00:26:44.069 --> 00:26:48.470 question about what you heard. You can go through 00:26:48.470 --> 00:26:52.140 it pretty readily that way, you know, pausing 00:26:52.140 --> 00:26:56.299 and going over it again as you would like. But 00:26:56.299 --> 00:26:59.259 having taught listening for so many years and 00:26:59.259 --> 00:27:05.619 also being naturally curious, I tried a number 00:27:05.619 --> 00:27:08.920 of tweaks on this. And this is where training 00:27:08.920 --> 00:27:11.759 myself, I didn't do this initially, but as I 00:27:11.759 --> 00:27:13.859 got into it more, I started doing it differently. 00:27:14.349 --> 00:27:16.990 So the first thing when you preview vocabulary, 00:27:17.609 --> 00:27:20.369 I hide the left side, which has the words in 00:27:20.369 --> 00:27:23.430 text, and just click on the audio button. If 00:27:23.430 --> 00:27:26.910 I can hear the word, understand it, that's what 00:27:26.910 --> 00:27:29.410 I want to do. I only go back and look at the 00:27:29.410 --> 00:27:34.710 text if I need to. Same thing with the responses, 00:27:35.470 --> 00:27:38.490 either type of activity with the responses. I 00:27:38.490 --> 00:27:43.890 hide the response options and listen. and then 00:27:43.890 --> 00:27:47.170 only reveal the response options once it's there. 00:27:47.470 --> 00:27:49.650 There should be a button for that, but there 00:27:49.650 --> 00:27:51.930 isn't. You have to just either avert your eyes 00:27:51.930 --> 00:27:57.009 or put your hand over it. I repeat if needed. 00:27:57.289 --> 00:28:01.230 I pay attention to the responses. I often listen 00:28:01.230 --> 00:28:04.309 again, but I don't go through this in some structured 00:28:04.309 --> 00:28:07.490 way every time. Harder ones I listen to more 00:28:07.490 --> 00:28:10.460 than the easier ones. And I... might even do 00:28:10.460 --> 00:28:13.079 some shadowing or take notes of useful words 00:28:13.079 --> 00:28:16.779 or phrases. One thing I really like about this, 00:28:16.799 --> 00:28:20.779 and this is the gen AI part writ large, is you 00:28:20.779 --> 00:28:24.220 can provide a new scenario and it will write 00:28:24.220 --> 00:28:29.059 new interactions for you. So I wanted to be able 00:28:29.059 --> 00:28:34.380 to talk about the weather for tomorrow because 00:28:34.380 --> 00:28:37.680 of plans. And that's all I said. Talk about the 00:28:37.680 --> 00:28:41.480 weather tomorrow. to make plans and it generated 00:28:41.480 --> 00:28:46.359 a picnic scenario and a making plans for travel 00:28:46.359 --> 00:28:51.720 scenario. Seems to be pretty good overall. There 00:28:51.720 --> 00:28:55.380 are some issues with authenticity. So in one 00:28:55.380 --> 00:28:57.980 interaction where I was buying a coffee, the 00:28:57.980 --> 00:29:00.720 phrase came up in French, how much do I owe you 00:29:00.720 --> 00:29:04.559 for your colleagues coffee? It's like, I'm checking 00:29:04.559 --> 00:29:09.140 out and I'm not buying. the service person's 00:29:09.140 --> 00:29:11.220 colleague's coffee, I'm buying my colleague's 00:29:11.220 --> 00:29:15.079 coffee. So it was a pronoun mess up there. But 00:29:15.079 --> 00:29:17.819 by and large, it's not bad. And when you notice 00:29:17.819 --> 00:29:20.640 things like that, it's useful as a learner actually. 00:29:20.799 --> 00:29:26.099 Mistakes can be valuable. So that's mine. That's 00:29:26.099 --> 00:29:28.900 a good note to finish on. Mistakes can be valuable. 00:29:30.220 --> 00:29:37.519 I think this actually also concludes our long 00:29:37.519 --> 00:29:43.460 discussion of our own paper. We're glad that 00:29:43.460 --> 00:29:48.240 we had the opportunity to introduce GEN .AI and 00:29:48.240 --> 00:29:51.400 SIPD, the Sustained Integrated Professional Development. 00:29:51.880 --> 00:29:55.359 This was the final part. We might come back. 00:29:55.460 --> 00:29:57.980 We've already entered that particular aspects 00:29:57.980 --> 00:30:02.680 like ethics and reflection were mentioned today. 00:30:05.439 --> 00:30:10.779 Otherwise, see you next time. All right. And 00:30:10.779 --> 00:30:21.960 see you next time, Mat. We are grateful for 00:30:21.960 --> 00:30:24.619 the support for opening AI for language learning 00:30:24.619 --> 00:30:27.279 by the Language and Applied Research Center at 00:30:27.279 --> 00:30:30.059 San Diego State University and the Southern Area 00:30:30.059 --> 00:30:33.009 International Languages Network. SAILN, which 00:30:33.009 --> 00:30:35.509 is part of the California World Languages Project. 00:30:36.130 --> 00:30:38.589 Mari Ocando Finol is the production coordinator 00:30:38.589 --> 00:30:41.829 of OAILL. Our theme music is by Tillmann Spiegl. 00:30:42.250 --> 00:30:45.329 Our editor is Chris Brown. Live conversations 00:30:45.329 --> 00:30:48.369 are moderated and promoted by me, Shahnaz Ahmadeian. 00:30:48.750 --> 00:30:49.710 Until next time.