Thoughts on (mostly generative) AI

This is a collection of talks, interviews, papers, and posts on the subject of generative AI, going back to 2018. I think the one I like most is my presentation for TAMK, that covers most of my concerns and a few of my hopes about genAI in relatively digestible format. Enjoy!

How AI Teaches Its Children: slides and reflections from my keynote for AISUMMIT-2024

My hopes and fears for generative AI in a 30-minute talk. Discusses ways to rethink education in the light of the risks posed by generative AI, not for obvious reasons like cheating or loss of teaching jobs, but because of how it affects what and how we learn. Two risks: the loss of skills (compensated for by the increased adjacent possibles) and how we learn ways of being through our interactions with LLMs (a much stickier and more profound problem). The title alludes to How Gertrude Teaches Her Children, by Pestalozzi, and, in particular, Pestalozzi’s erroneous belief that we can scientifically establish teaching methods that, if applied, will inevitably lead to effective learning.

 

Sets, nets and groups revisited

Rethinking the nature of groups and collectives in online learning, with some observations about potential roles for generative AI.

Proctored exams have fallen to generative AI

An example of the use of generative AI in a proctored, in-person exam. This particular attempt failed because of the student’s behaviour, that I reckon to be a failure in technique, not in the tech used to cheat. Others will be doing this more successfully.

Interview with David Webster

An hour of thought-provoking chat with Dave Webster about generative AI

▶ I got air: interview with Terry Greene

Interview with me about my book, How Education Works, with the tail end of the conversation about generative AI

Educational ends and means: McNamara’s Fallacy and the coming robot apocalypse (presentation for TAMK)

Includes slides and video. Talking about the problems that proxy measurements of learning that work for human-based learning really really don’t work well when AIs are involved.

Stories that matter and stories that don’t: some thoughts on appropriate teaching roles for generative AIs

Do we really want AIs telling stories to our kids, and what do they learn as a result? Some thoughts.

Presentation – Generative AIs in Learning & Teaching: the Case Against

A Lunch n Learn session for Athabasca University, with a brief to present a negative view of generative AI.

Published in Digital – The Human Nature of Generative AIs and the Technological Nature of Humanity: Implications for Education

The most rigorous paper I have written so far about my fears and hopes for generative AI, published in Digital. It’s not perfect, but it does have references!

Recording and slides from my ESET 2023 keynote: Artificial humanity and human artificiality

A rabble rousing discussion of the risks of generative AIs

10 minute chats on Generative AI – a great series, now including an interview with me

Three short commentaries around three talking points, ably guided by Tim Fawns.

Cognitive prostheses and the future of the human race

Doom, gloom, and the slow fizzling of human intelligence.

The artificial curriculum

Thoughts on the hidden and tacit curriculum, and the abject failure of generative AI to meet the non-explicit aims of education.

CHATGPT DOESN’T ALWAYS GET IT RIGHT, BUT THE FUTURE OF AI IS ‘SO SCARY AND SO EXCITING,’ SAYS AU PROF

A report on an interview with me, not exactly representing my views in all their complexity, but a good news story from early in the ChatGPT pandemic.

View of Speculative Futures on ChatGPT and Generative Artificial Intelligence (AI): A Collective Reflection from the Educational Landscape

Great paper on AI possibilities – I am one of 36 co-authors. This one is being cited at a phenomenal rate that I have never experienced for any previous paper.

Two stories about learning to be human from a machine

Comparing how ChatGPT and I might go about writing the same story about being taught by generative AI.

Loab is showing us the unimaginable future of artificial intelligence – ABC News

Scary, spooky, wonderful. Pre-ChatGPT.

So, this is a thing

Pre-ChatGPT discussion of what happens when both students and teachers turn out to be AIs.

Can GPT-3 write an academic paper on itself, with minimal human input?

Pre-ChatGPT post raising an alarm bell or two about where, just a few months later, we actually found ourselves.

A modest proposal for improving exam invigilation

Some thoughts on using robots as remote proctors. Possibly a little satirical in intent.

At last, a serious use for AI: Brickit

Reporting on a wonderful app for suggesting Lego models to build, based on available bricks identified by a discriminative AI.

Smart learning environments, and not so smart learning environments: a systems view | Smart Learning Environments | Full Text

2018 paper making the point that smartness is a feature of the entire technological assembly, not its parts. AIs are only parts.

AIs can pass SATs. So, what does this tell us about SATs?

Brief post from 2015 noting that generative AIs had, at that point, already reached the point they could pass a SAT with an average mark. We’ve had a few years to fix this problem and yet we have made very little progress doing so.