Unrealia

I am used to the rapid changes in technology resulting in frequent decisions to change tools, migrate from one platform or tool to something else. This is part of the landscape that educators and learning technologists inhabit and, to some extent, it is something we accept. It is the reason for highlighting the importance of critical thinking in technological choices and for being an active member of a community of practice such as the Association for Learning Technology

As an educator however I am finding some of the recent developments and hype even more challenging that usual. Having knitted great connections through the platform formerly known as Twitter I now encourage my networks to move elsewhere for their interactions. The platform is no longer what it was, the openness has gone and users are even more vulnerable to exploitation through financial demands (eg. for the meaningless blue tick) and transparency available through aggregation using community tools has been lost. I remain present on X for the time being but meanwhile I am developing my networks on Bluesky and Mastedon and recommending professional organisational activity move to LinkedIn and, if appropriate Insta.

From a language education point of view I am taking great critical care in monitoring the development of Artifical Intelligence and, most specifically Generative AI such as ChatGPT and AI powered image and video generators. This recent post on X caught my eye:

If you know any Spanish then I dare you to translate these AI generated “unrealia”. Realia has long been a vital part of communicative language teaching and, once access to the internet became more readily available, I celebrated how it made it easy to bridge the distance between French culture and the experience of my students based in the UK. I could easily link to real restaurant menus and information from the milenium onwards and I would teach my students how to navigate such realia. Using an AI generator to create such resources solves a non existent problem and creates many new ones! I congratulate Mr E on demonstrating this so clearly.

I am also concerned about the ease with which images, video and voice can be faked. Granted there may be opportunities here. One submitted abstract I reviewed recently pointed to the value of being able to create a video of yourself and (using AI) show you speaking fluently in a language you do not actually speak. I hope to see that presentation shared in the EUROCALL annual conference at Trnava, Slovakia in August. Whilst it is true that language learners often struggle to imagine themselves as a fluent speaker of a language they are learning and therefore such fakery could provide psychological encouragement, the risks of using such technology are as yet poorly understood. I have always worked for the improvement of voice over the internet but in the case of such new tools I want to know what is happening to the recording of your voice when you submit it? Voice recordings are increasingly significant data which can be manipulated and used for nefarious purposes.

By far the questions which need to be asked remain unspoken in the light of the flashy novelty of creation and the anticipation of reducing time and expense. Here are some of mine:

  • Large language models have to be trained : we need to know what material they are trained upon as this will always influence what they generate
  • Who benefits from your access to ChatGPT or similar AI generators. I am sure it is not just you!
  • Who is likely to lose out as a result of applying AI to your workflow? Artists? Creators? Teachers? How can we trust what we see?
  • When there is so much in fighting in the commercial sphere of these new start ups should we not be concerned that pace of change is making it impossible to really understand the risks?

As an open practitioner I really value the conversations going on in spaces such as OE Global Connect. Anyone can join and tap in to a good deal of technical expertise. Learning together is both fun and important if we are to have a voice in this rapidly changing landscape.

2 responses to “Unrealia

  1. We have this so called curse of living in interesting times, eh? I respect your efforts to bust out of “X” and I think the idea what was effective there from the early days will ever be recreated in a single space. It used be said by a smart author that the future was here but unevenly distributed, but now it’s the present as well?

    Also, thanks for mentioning/linking to OEG Connect, and hence I am torn between commenting here and there or both.

    Your Four Questions are valid, important, yet I feel rather beyond our control. Generative AI is hardly a thing most of us would have chosen as a new path (going to the web was more a choice in the early days, not a clobber over the head), but also we are stuck in a mode of being reactive since the platforms and tools are operated behind walls.

    Sadly I think the answers are:

    • We likely will never know the training sources, and in fact we are seeing developments of “synthetic” training which is some kind of mixing up everything as more goo. Results from LLMs are mathematically impossible to trace back to specific training data.
    • Of course the providers benefit first, we are just in their Matrix.
    • Many will lose out, look at the shredding of the news industry, hard to see how education, content creators are going to be immune. Trust what we see? Everything ought to be suspect, this is the new age!
    • Alas, you have nailed it, not only the pace of change, but the sheer shortfall of our minds and concepts to really understand systems operating in dimension spaces of 100s. The results are not deterministic and usually not replicable, so all ew do is base our intuitions on what the machine spit back.

    Keep asking and talking everywhere is about all I have. And also, at many points, ask yourself, is this the best way for me to accomplish a task or need? Is the convenience and allure of making something quick and on the thin surface sufficient, really better than trusting what we create ourselves?

    Carry on, and as a wise man now heading to retirement says… “Blog!”

    • Wonderful to have your feedback on my blog post Alan, and I fully agree with you. It feels like the world of tech runs outside and with little awareness of the impoartance of the “common good” and as govs also fail to understand how to control or regulate it falls to our communities to be well informed and involved. We are better together!

Leave a reply to CogDog Cancel reply