Why AI alone can’t help our children learn: A personal account

Not long ago, I was emptying out one of my twin kindergartner’s backpacks when I saw an unusual package from his school. It had my son’s name, a series of passcodes and a download code.

I was puzzled, given that my husband and I had been assured kindergartners didn’t need computers. But now, thanks to one AI-based test company’s teaching tools, my children would be using an app to — supposedly — learn completely on their own, at their own pace, with no help from us.

I’ve spent decades researching misconceptions about the nature of intelligence. So, while I was curious as a parent, I was also interested as a scientist to see how this would play out: Could AI really teach my kids?

I am also a professor at a large public university where budgets are tight. How AI will transform education is of incredible import to schools such as mine, teachers such as me, and hard-working caregivers such as the parents in my community. Could AI offset our struggle to get our children to grasp new concepts and skills? Might AI be better equipped to help them tap into their own intelligence?

After a few days watching my bouncy twins scroll and click, I can tell you the short answer is no.

I should note that I am by no means anti-AI. Our home is already awash in it. We have Alexa. Everyone in our family uses AI programs in their work. And we often let all three of our children, the kindergartners and their toddler brother, do their own projects with our tools.

After OpenAI introduced Dall-E in 2021, Grandpa helped the kids create artwork mashing up their favourite images and themes. When ChatGPT launched last year, my husband let the children use it to do deep dives into yetis and cephalopods.

Not sceptical of the ability to have fun

But the more I watched my children, the more it became clear to me that, while AI can assist in getting information to a learner, it cannot do the thinking for them — it cannot help them truly learn.

You see, human intelligence is different from computational intelligence. For one, human intelligence is not quantifiable. The brain is plastic, always developing and growing as we learn from our environments.

In addition, human intelligence relies on human interaction. Exercising our intelligence is something we do naturally as we connect with others, consider the world around us and seek to improve our relationship to that world.

We all share this drive to know and connect. But to tap into this most basic birthright, we must follow the learner’s natural curiosity and passion. Without that social-emotional piece, information is merely tallied and later lost to other pursuits.

In education, there has been a movement toward social-emotional and problem-based learning as a springboard for academics — and with good reason. Research shows that students across age groups and skill levels learn better and retain information longer when they are offered the opportunity to see how the material they’re learning connects to their lives outside the classroom.

So educators attempt to spur interest in subjects and acquisition of skills by turning learning moments into communal problem-solving events. When we combine analytical learning with social-emotional learning, students become proficient in the material we want them to know and get more excited about the learning process.

Student uses of AI 

AI learning often involves an individual working alone with a bot. The bot does the research to, as one AI tool says, “get you instant answers.” It can crowdsource information to help students find facts about their environment, solve a problem and come up with a creative way forward.

But AI doesn’t compel students to think through or retain anything. And simply being fed facts and information is not the same as “learning.”

Ultimately, if you want students to learn, they need to shore up their neural networks and use their neuroplasticity to develop their own intelligence. This is where AI falls short. There is nothing better than collaboration in real life — connected, reciprocal learning between a student and their peers or teachers — to spark the brain’s natural drive to develop and grow.

When my kids engage with AI, the interaction inevitably fizzles out. Eventually, they need to move their bodies, look one another in the eyes and communicate as they tackle a new skill.

By contrast, they can read or do math with their little brother for long stretches, even longer when they are teaching him — when it’s human-to-human, analogue, with love.

My students are using AI with even less success. They simply ask ChatGPT to answer their assignment questions. They then get passed through the class with top scores. Will they get a degree fast? Sure. Will anyone doing this retain the course information? Unlikely.

It is enticing to imagine that a bot might cure what ails us. In state school systems like the one where my children are taught and I teach, there are chronic staffing shortages. Many parents don’t have the resources — in time, money or energy — to teach their kids at home. If a bot could fill the gaps ... how great, right?

Yet the atomised nature of AI “teaching” as it currently exists means that students merely level up without learning. When it comes to cultivating intelligence, nothing can beat what we humans have been doing, face to face, for centuries.

Washington Post

Rina Bliss is a professor of sociology at Rutgers University and the author of “Rethinking Intelligence: A Radical New Understanding of Our Human Potential.”