"Useless for learning" is just wrong. I've found LLMs immensely useful for directing my learning projects. Of course, a lot of the actual learning must come from doing things and puzzling through them myself. But I now find LLMs to be indispensable in finding out what I need to learn to accomplish a task, finding keywords to search on Wikipedia or in textbooks, and answering questions when I'm confused about something.

I'm an adult with a fairly balanced view of AI and I find it difficult to learn coding without occasionally using AI to help me navigate to the most relevant bits of MDN or help me check if my thinking on an approach is correct (it's all entry level stuff so should be well represented in the training data).

I find it easy to to into a long chat with an LLM about some project I'd like to try and what's involved with it. I find it easy to get into a chat with an LLM about a lot of things as a kind of unproductive excursion that my brain tells me at the time is 'useful'. I'm of average will, so I dread to think how this will work out with children who get to 'partner up with AI to assist them' or whatever marketing speak is used to obfuscate their goals. Then combine that with social developmental issues or below average focus.

It's bleak because the more entangled they get with the system the more they'll seek to push back regulations.

Using AI in an intentional way with purpose and direction should be great for practicing thinking.

The right way to teach children to use AI is to teach them to scope, filter, design process, edit, refine. How to ask a question, how to think through steps, how to use language to describe all these things. Each kid has something that can think and respond as fast as they input.

The goal should be to perfect sequence and iteration, not skip to final output.

These skills also should NOT be framed in some kind of "teaching AI" as much as teaching communication and critical thinking and analysis. It is the exact same skills you need to solve problems and interact with humans.

> Yeah, I cannot imagine how anyone could learn anything well with access to AI.

You must not have much of an imagination then. Or maybe you're just being overdramatic? AI is arguably the best way to learn most subjects now. Frontier models have made a lot of progress on reducing hallucinations, and AI can teach you at whatever pace you're capable of learning it. There are very few topics it can't teach, and it can go into more depth than you'll find in any textbook.

I would not say that. My child asks the AI factual questions the same way she would ask an adult. That's one kind of learning. There are others, of course.

The missing bit is a representation of knowledge, and a way to represent a learner’s comprehension.

Even if you shortcut by synthesizing a textbook in every major topic - that’s just one arbitrary representation, and the way topics overlap is outside of the material.

I am very interested in this though, if anyone has references to relevant research I’m all ears.

"You're absolutely right, Nell. I shouldn't have confused ethylene glycol with propylene glycol. Would you like to know more about funeral services?"

If you care about the information and communication, and you think you can do a good job of the slide deck if you think through it for this venue and audience -- and maybe even have new insights by going through the process -- then it can be enjoyable/rewarding.

But I've also seen situations in which the presenter doesn't care, or the slides are just a backdrop for some better communication/selling/maneuvering they're doing, or they know the information is bogus or the presentation pointless, or they know the audience doesn't care, or for everyone it's just a meeting to be able to say you had a meeting.

I'd guess that at least half the current use of LLMs is for "cheating on your homework" tasks, in which the person prompting it simply doesn't care -- whether it's for schoolwork, professional work, or socializing.

David Byrne seems to like it: https://newsarchive.berkeley.edu/news/media/releases/2005/03...

I haven't seen his actual slide deck anywhere online though.

There was one I did have fun making:

Back around 2007 it was about AI, and for part of it I'd memorized like 2-3 minutes of what I was going to say, along with careful timing. The plan was that I'd start wandering around, including around the back of my laptop so I wasn't looking at the projector or laptop screen - and during that time, robot characters would wander onto the screen and start running around. The idea was I wouldn't look back until right after they hid themselves. I think I'd even put a spot in the middle where I could glance back while they were hidden, then they'd come back out.

I don't remember why, but I never got to present it.

There are two parts of making a slide show: 1. Visualizing what you want to show 2. Finding tools to show it

Developing 1 means you need to start with pen and paper without any pollution from existing tools. You might read and experiment beforehand and use as many tools as possible (including AI) but at least once before you make thr final version you should sit down and just think from scratch what you want to show.

Then you can use AI for 2. Teaching kids 1 is very difficult while simultaneously giving them access to AI since they are too young to develop that self control.

Back when I created them (high school), I enjoyed it, because it was about making an appealing presentation about the data we researched.

It's never been about making education better. It's about building and legally mandating a money hose directly from your town property tax levy to accounts receivable at Google and Microsoft and OpenAI.