LANGUAGES, WOOD & TREES

I first read Sun Tzu’s Art of War over 30 years ago, when I was just starting out on my path to become a teacher and researcher of languages. All I knew then about AI what I had read in Sci-fi, and I still feel that Sci-fi must be part of the discussion whenever the implications of paradigm-shifting technology are concerned. Most of us simply don’t have the imagination to know where it will take us…

Sun Tzu said “Know the enemy and know yourself in a hundred battles you will never be in peril”. Of course, language teachers have always been early up-takers of disruptive media-based technologies – I mean let’s face it, we need all the help we can get. I mainly focus on pronunciation in my work, which now includes teaching, research, and teacher-training, and I have always chosen to make digital technologies and autonomous learning a lot of time in all of these aspects of my work. When I teach in the classroom, my approach is very physical: a combined approach of both embodied and cognitive / meta-cognitive techniques. But I have always kept the digital enemy close to me. In fact, I would go so far as to say that digital technologies are more of a friend now. 

I’m not yet sure about AI, perhaps it’s more of a frenemy. What I am sure of is that most stake-holders can’t see the wood for the trees. To be honest, most of the decision-makers in French education have yet to even see the trees… AI is a reality and has been for many years. Most of my colleagues, myself included, have been using Large Language Models (LLM) such as DeepL for translation for years, and it certainly saves time. In our research, some of us use ChatGPT and other LLM tools to write first drafts of reports, grant proposals, abstracts and even articles. But even if individuals are often quick to find uses for new technology in education, education as a whole is often slow to adapt. In our teaching, we see our students using LLM tools more than we would like, and we are struggling to find suitable ways to assess what they are actually learning. 

Those of us who use LLM tools know that the devil is in the instructions we give the tool, and in the editing process. Obviously, we have two choices here – ban AI use in some or all learning and assessment – which seems to me like a King Canutian solution. Or lean into the tech and get on with the business of learning, teaching and assessing. I have seen some interesting charters for AI use in education, many adopting a scale from 0 – 4 when assessing students’ work, i.e, all parties agree that, for example, 0 = “no AI was used in this assignment” to 3 = “this work is entirely the product of AI”. As yet, I have received no guidelines either from my faculty, my university or the Ministry for Education here in France. 

So much for the trees. The wood was nicely summed up by a meme I saw a while back: in the first picture, the guy was smug, and the caption read “me, realising AI can do my work”, and in the second picture, the same guy was horrified, and the caption was identical. And there’s the rub. If AI can do your work, AI can replace you. Or at least many of you, maybe even most of you. There will always, I believe, be a role for teachers, but if we want to remain relevant as the number of required human teachers for the number of human learners decreases, we had better heed Sun Tzu’s advice on knowing our enemy. And perhaps on knowing ourselves, too – but that’s a much bigger subject…

-Dan Frost, associate professor, Université Grenoble Alpes, France