So says Dr Michael Johnston, senior fellow at The New Zealand Initiative, who has authored a new report outlining both the dire risks and tremendous opportunities generative AI presents to education. 

Johnston is critical of grand claims that AI will dislodge the need for students to learn writing skills.

He says the notion that the technology will free up the new generation from such tedious learning processes, and instead allow them to think more critically and creatively about AI outputs, is deeply flawed. 

“This is one of the arguments that was advanced in a webinar that I listened to in preparing the report, and I read elsewhere as well – it’s kind of a ‘utopian 21st century learning thinking’,” the expert tells EducationHQ

These sweeping proposals most often don’t come from educators themselves, Johnston says, but from enthused tech entrepreneurs naturally keen to push their offerings in the sector. 

“You can understand why they’re interested in promoting the idea that AI will fundamentally change education – it’s a bit like the big tech companies being quite pleased to put computers on every child’s desk; they’ve got a product to sell… there’s a real risk there,” he says. 

The expert contends that writing is “one of the greatest technologies for the enhancement of thinking ever invented”, and says when we consider cognitive load theory and the science of learning, generative AI tools do not remove the need for students to acquire writing skills for themselves – even if the likes of ChatGPT can compose more sophisticated prose in a fraction of the time. 

“Let’s think about writing, it’s a powerful means of communication. But not only that, it’s a tool of thinking,” Johnston says. 

“Because when we write about something, we’ve developed fluency with the basic skills of writing, and then we’re able to use writing to record our thoughts while we’re developing them. 

“And that actually reduces the load on our working memory … so if young people don’t learn fluency with the basic skills of writing, they won’t be able to use writing for that more powerful purpose.”

Students’ success – both at school and in life – has been put at risk by similarly misguided thinking in the past, the expert notes. 

He recalls the debate that swirled over the need to learn foundational maths skills when calculators became widely available in the late 1970s. 

“Now we [had] this technology that was easily accessible, that could do it for them. Why would they need to go through all that difficult effort of learning arithmetic? 

“The answer is that if they don’t, then they’re not able to access more advanced mathematics,” Johnston says. 

If, for example, students do not become fluent with addition – where the skill is safely encoded in their long-term memory – they simply won’t be able to access or indeed master more advanced mathematics, he adds. 

“Addition will still have to happen using their working memory resources, which will make it nearly impossible for them to learn about multiplication, which depends on the knowledge of addition. 

“And mathematics is a very hierarchical subject, meaning it’s layer on layer of concepts. And so, each layer needs to be securely enclosed in the long-term memory before we attempt to build on it.” 

The reason so many people feel they are rubbish at maths is because this structured teaching approach doesn’t happen all too often, he says. 

“It’s not that they’re not good at mathematics, it’s that they were taught in a way that didn’t ensure that each stage was securely encoded…” 

AI’s capacity to potentially offer one-to-one tutoring on a grand scale is a huge positive, Johnston says.

In order to think critically and creatively, students must have sound disciplinary knowledge, Johnston points out. Without this, “thinking is empty” he suggests. 

It’s a point raised by prominent education psychologist John Sweller, who has previously contended that there’s no single strategy that can be used to teach students how to think critically or creatively. 

And in Johnston’s view, the best safeguard against the spawning online misinformation we’re seeing comes down to instilling students with a lot of knowledge. 

“My argument is that it’s not censorship, forget trying to shut down misinformation, I don’t think it’s possible,” he says.

“And even if it was, the best protection is to arm young people with a lot of knowledge – a lot of critical thinking capacity rests on having knowledge, and then they are able to navigate the world better, they’re able to kind of sniff out bad information and wrong information and when people are trying to deceive them, or even advancing bad arguments in good faith. 

“They need to be able to think through those things for themselves. And to do that they need a lot of knowledge.”

The same applies for creativity, Johnston says. 

“You can give a toddler a crayon … but to be truly creative, they need to learn some techniques. 

“They need to learn how to paint well or draw well. [In music] any child can bang on a piano but it’s not the same thing as being a musician. 

“…to be really creative with anything we need skills and knowledge.”

Despite the bad educational ideas that have resurfaced with the rapid advancement of AI, the expert says school leaders can harness its potential to improve teaching and learning in ways not previously possible. 

There are many positives ripe for the picking, he suggests. One is AI’s capacity to offer 1:1 tutoring on a grand scale. 

“Paul Bloom was an educational thinker back in the late 20th century, and he talked about the ‘2-sigma problem’,” Johnston explains.

“So, some research had shown that if young people are given one-on-one tutoring, then the average person will achieve about two standard deviations above the average student not given [it].

“Two standard deviations is massive. It actually takes you from the 50th percentile to the 95th percentile…”

The obvious problem is that we can’t afford a one-to-one tutor for every child, Johnston says, but the advent of AI could potentially change this if used responsibility. 

“And by responsibly, I mean under the supervision of a skilled and knowledgeable teacher, not just a child left with an AI to do whatever they want.

“And not just any AI, but one that’s been purpose-designed to support education,” the expert says.