ChatGPT 6 months on. Blogjune 5/23


It’s minimalist Blogjune – one pic, one thing I did today.

What I did today: Attended a two hour online seminar along with 3000 other people. Hosted by the higher education regulator, TEQSA, it featured a panel of experts discussing what the sector has learned after our first semester teaching in a world with ChatGPT in it.

My takeaways:

  • Every expert or commentator seems to have their own take on the significance and impact of generative Artificial Intelligence and how the sector should respond.

    I have lived through many tech changes and the big difference to me now is the lack of consensus among people who I would expect to be well-versed. I do not remember such diverging views in past changes like internet connectivity, wide-use of Google, Wikipedia or social media. For these changes I usually found someone whose opinion I trusted and agreed with.

  • Assessment must change. Yesterday.
  • This is not necessarily due to student access to AI. What passes for assessment is a shoe-horning into courses of easily countable activities that can be marked in one hour per student per semester, whether the tasks are relevant and authentic or not. Resourcing for good assessment left the building over a decade ago and much of the problem of assessments easily cracked by AI is due to assessment design that is basically an AI chatbot prompt. No wonder AI chatbots can do them well enough.

  • Currently comfortable and secure knowledge and communications-based university subjects will become obsolete quickly.
  • I suspect my own discipline, librarianship, may be on the list. I teach an information retrieval course, which I suspect demonstrates increasingly artisanal skills for resources rapidly becoming obsolete. For a few years I comforted myself that our profession still had the essential human skills of helping clients refine their questions, recommending the best sources and helping clients evaluate the authority of information. When probabilistic word strings become acceptable as knowledge and “good enough” is the acceptable information standard, I am not sure that knowing how to manually navigate knowledge systems that produce outcomes differently and far,far more slowly, will be a valued skill.

  • As an educator, I am not so concerned about ChatGPT. It’s an easy to spot target with enough eyeballs and clucking tongues dealing with it.
  • What keeps me up at night is the AI-enhancement of tools students have been using for years. I already had problems with Grammarly as a “writing assistant” that homogenises expression and actually prevents skills from developing. Not everyone agreed. Now that it has been enhanced to become basically a predictive knowledge base, some people are digging down because it is familiar. And Microsoft Co-Pilot? It’s here with predictive text whenever you add words to an online interface accessed with Edge web browser. Like an online forum that is being assessed. Or an online test.

2 thoughts on “ChatGPT 6 months on. Blogjune 5/23

  1. Sounds like an interesting workshop. I’ve never been good at predicting trends (though I was right when I told workshop attendees in 1998 that Google was going to be a big thing… 😂) But I’m seeing AI as both as a big change and at the same time, just another piece to incorporate into our knowledge management. information/media literacy/detection skills become all the more important.

What do you think? Let us know.