Duration: 5 minutes 39 seconds

[music]

Adam Spencer

I've been a numbers nerd from the youngest of ages so maths was always the most natural of all the languages I encountered. And the further we go into this digital century, the more maths and your basic STEM skills are gonna be the ones that build the years out in front of us.

I think AI is gonna change education as much as it changes any industry. It's gonna change everything about it. The way students learn, you'll have a personal tutor that stays with you for life, from primary through high school, TAFE University, then through your professional career, it'll know how you learn, what you need to know, the best way to teach it, and students will use these things more and more, hopefully not to avoid critical thinking or just cheat but to turbocharge the capacity their critical thinking skills give them to interact with the world.

From my days of mathematics, the old idea of have an exam, five simple questions, five medium questions, five hard questions, then the five questions I used to love doing at the end, the really tough ones. Well, if a student gets the first three easy questions wrong, there's no point going to the medium and the hard stuff, let's stay there and work out what they really do and don't know, where are the blockages. If someone hits the first three easy questions out of the park for six in two seconds each, do they really need to do the medium stuff? No, let's dig down here and see how good they are. Deep, rich feedback on individual students.

The days where everyone in this class is doing the same level of geography because you're all between 14 years, three months and 15 years, nine months old, that makes no sense at all, in an age of truly interactive large language models and things like that, it's gonna change in ways that we've never seen before.

You have to push back against this, let's just ban it, let's just keep it away, we have to embrace it from the earliest possible stages. Kids should be using these things at home, not to cheat and do the work for them, but to explain, in a preliminary sense, something before they're engaging with the harder aspect of it.

My daughter's just started at university, she's doing politics. She finds politics really hard, she's never studied it before. So she said to the lecturer "Should I go to your lecture first, then do the readings 'cause I'm finding the readings tough, and then do the tute, or should I do the readings first before the lecture?" Lecturer said, "Come to my lecture, then do the readings." What she does now is loads the reading into ChatGPT, says, "Can you explain this to me as though I'm a smart 16 year old who knows nothing about the subject area in a thousand words, please?" It gives her a little primer as to what the reading's about. Then you go to the lecture, then you do the full strength reading, and then you're hitting it, you're absolutely ready to go.

So you use these things to distil and summarise information for you and pitch it back in a way you'll understand. Now she's cheating if she just hands that in as an essay or if she doesn't then do the full readings, but as an introduction, as a way in, it's totally changed the subject for her. The way you ask the question affects what you get back from a large language model. If you ask the question a few times, you'll get a different, hopefully better answer.

And you have to be careful and check things 'cause they do make mistakes. If we bed that into kids from the start, the earlier we get them using these things, the better, because they're all gonna go into high school and universities and they're gonna go into careers where these things are being used. The idea that you'd say to a law student at uni at the moment, "You can't use ChatGPT," if I was that law student, I'd say, "Dude, in two years, I'm clearly gonna be using it at the law firm I'm at."

The mobile phone revolution didn't happen instantly. It was a few generations in when you could download video en mass and that sort of stuff, where we realised, wow, this is totally game changing. Whereas large language models, when you went from GPT-3.5 to 4, there was an instant giant step change in the amount of grunt these things had and what they could do.

Everyone's talking about it right now because the magnitude of that jump, that's both exciting because, for once, you've got a generation of kids who are switched onto a technology and wanna know more about it and wanna learn about it. It's daunting because, with the greatest of respect, education doesn't always move rapidly as an institution and as a portfolio. And the challenge is now to do so because the kids are gonna get it. And if schools take five years to catch up to the fact that kids are realising, well, hold up, this thing could pass the US Olympiad biology exam if it wanted to. And if we let the kids get ahead and just be really good at using GPT and we lose a generation of students from critical thinking and analysis and forming their own ideas and expressing 'em, that's a generation lost. So it's a tremendous opportunity but also a real challenge for education to keep up.

Students have to realise the potential of STEM subjects right now is that more and more careers are gonna explode in those places. Some jobs and types of jobs and careers that don't even exist yet, but the STEM fields are just gonna continue to open up. The most important challenge for education at the moment, I think, is to get teachers who are passionate and qualified to teach those subjects. Telling the message that this is where jobs are gonna come from, this is where wealth's gonna come from, this is where nations are gonna be built. It's a STEM future.

[End of transcript]