The opportunity of AI: How ChatGPT can revitalize US education

If we’re asking children to write summaries, ChatGPT is an incredible way to cheat.

So goes the argument in Will Douglas Heaven’s recent article in MIT Technology Review. And it’s true — as a speech pathologist and Harvard lecturer, I’ve worked with many kids who struggle to read and write and who would love to have a computer do their writing and thinking for them. But as easy as it is to say that ChatGPT signals the end of education, the real story is far from that simple.

Already many articles are discussing how we can use ChatGPT and other large language models to improve education. But we can actually do something even more important: We can use this new revolution to rethink how we envision success for our children.

Instead of helping children get ahead in a single-minded, robotic way, we must think about education and doing a deep dive into our — and their — humanity. We can ask them to connect their knowledge and their values, to think aloud about what interests and excites them and what they fear or are worried about. We can assess them in ways that recognize their unique and developing thinking skills.

We also must think about school differently, as a place to integrate heart and mind, not simply rehash or even analyze information. In fact, our worries that kids can cheat on summarizing activities is a symptom of the problem. Why, in our age of constant Googling, are we focusing on children’s abilities to simply rehash what they have read?

We do know that summarizing is an important way to support reading comprehension, but if we stop there, we leave children without any understanding of why this information matters or what the purpose of their reading was. This problem worsens when we think about how we often introduce reading materials to kids. So often, we have kids read one passage or news article, then answer questions about it, then move to another one. One article or passage isn’t necessarily connected to the last. And most likely, neither of them are connected to children’s real lives.

When kids say they’re not interested in reading, even if they have the skills, it’s likely because we haven’t tapped into their intrinsic interests or needs. I’ve seen this with my 6-year-old son — he has above-average reading skills, but when asked to focus on reading simple narratives, he often shies away. In contrast, when allowed to choose books about species of snakes or sharks, his favorite animals, he brings books to the dinner table and often asks, “is this the right way to read this?” or “Hold on, I didn’t get this word.” Even with books far above his reading level, he wants to engage and give it a try.

Other kids may feel the same way about graphic novels, or about books that explore deep sea diving or mountain climbing. If we want to make kids increasingly robotic, we can continue focusing on giving them all the same books. We can even focus on doing slightly better, and differentiating books based on children’s reading levels. But all this doesn’t go to the real purpose of helping children read or learn. That is, we want children to develop their full humanity and to celebrate and understand the humanity of others. We want children who behave less like computers and more like themselves.

Raising children in this way requires a full re-envisioning of the way we understand success for children. Such a vision involves thinking much more about children’s abilities to judge the ethics of a situation, to apply empathy in difficult times, and to explore the short- and long-term consequences of actions. It requires helping children know what it means to have compassion, and how to recognize and support well-being and others and themselves. It means helping them develop the skills to handle complex generational challenges such as the environmental crisis and, ironically, the benefits and dangers of AI. And, most fundamentally, it involves helping them know what it means to love and care deeply and authentically for others, and to know and love their authentic selves. Instead of diving further into the airbrushed world of Instagram images, we need to support them to love and care for their authentic realities.

In school and at home, we will also need to assess them differently. We will need to focus on the quality of our conversations, in terms of how they support empathy, compassion and critical thinking. We need to emphasize the everyday back and forth of our discussions, as they reveal both how much children understand and how they feel. We likely need to get away from data-driven metrics such as “How often does this child raise his/her hand?” and toward reflective listening that focuses on what children are trying to communicate.

To be clear, it’s not that we shouldn’t notice if children are raising their hands — we should notice, just as we should notice if they’re absent from class. We know that hand-raising and talking do give information about how engaged students are. We also know that it’s critical to support all children in feeling welcome to contribute and have their voices heard, especially those from marginalized communities or whose families have been historically silenced. The problem comes when we over-focus on metrics like hand-raising and number of comments. Over time, we may come to see these metrics as a substitute for the quality of learning taking place. We may even come to see children as data points — notable for their inputs and outputs — rather than as developing, relational beings.

With their holistic development in mind, we may well use ChatGPT and similar programs to support students’ learning at all ages. In fact, we would be remiss not to. For instance, we can ask children to generate multiple solutions to a problem by using ChatGPT, and then evaluate the consequences of these solutions and how well the arguments are being made. More fundamentally, we can ask about how well their values and feelings align with these solutions. “What worries you, when you read about that?” or “Who do you fear could be injured by such an argument?” When reviewing summaries from ChatGPT, students can analyze which elements the summary is focused on and which elements it may have left out.

In the words of Will Douglas Heaven, the writer of the article mentioned above, “ChatGPT is going to change education, not destroy it.” We can take a more radical stance. We must change our beliefs about education and success in this brave new world. At the risk of sounding hyperbolic, our children’s humanity depends on it.

Rebecca Rolland is a speech language pathologist and adjunct lecturer on education at Harvard University. She is the author of “The Art of Talking with Children.”