Answer extracted from the The Diary Of A CEO with Steven Bartlett podcast — listen to the full episode below.
What concerns exist regarding AI use in early childhood brain development, particularly for children ages 0 to 8?
The concerns are profound: one-on-one human interaction with a skilled teacher remains the scientifically proven gold standard for early childhood brain development, yet it's labor-intensive and expensive. While AI could theoretically scale this ideal, the real problem is far deeper—we don't know whose values, biases, and cultural priorities are being embedded into these systems. The question isn't whether AI *can* teach, but whether we can ensure it teaches the *right* things.
The neuroscience here is clear-cut. The human brain in early childhood—especially ages 0 to 8—develops through relational interaction: a caregiver who responds to a child's needs, adjusts their teaching, and embodies the values the community wants to pass on. This isn't negotiable; it's how neural pathways form.
But there's a scaling problem. One-on-one teaching at scale is economically impossible in most parts of the world. This is where AI enters the conversation—as a potential equalizer. If you could train an AI system to teach the way a great teacher teaches, adapting to each child's needs and learning style, you could theoretically reach millions of children who otherwise have no access to quality early education.
The catch is the values problem. As the neuroscientist explains in the episode, these AI systems are inherently biased. They reflect the cultural assumptions, moral frameworks, and priorities of whoever trained them. If an AI is trained on Western individualistic data, it will encode those values. If you're a parent in a culture that prioritizes collectivism, interdependence, or different educational outcomes, that AI may actively work against your family's values without you realizing it.
The Hidden Values Problem
This is less about the mechanics of AI and more about power. Whoever designs and trains the AI controls the moral and cultural curriculum of millions of children. That's a staggering responsibility—and it's largely unexamined in most AI development today.
The risk isn't that the AI is "bad" in an obvious way. It's subtler: a child learns from an AI that encodes values misaligned with their family or culture, year after year. By age 8, their neural architecture—their way of thinking, their sense of right and wrong—may be shaped by someone else's worldview.
There's a deeper concern embedded here: as discussed in the podcast, nearly 30% of US parents with children ages 0 to 8 reported their kids are already using AI for learning as of March 2025. Most of these parents probably haven't thought about what values their child's AI is imparting.
"You have to have a relationship with it or it's going to turn toxic. It's going to hurt you. But if you have a good relationship with it, it can make your life better."
Unknown Unknown — Neuroscientist and child psychiatrist; president of the foundation running the Neural Information Processing Systems (NIPS) meeting, which attracted 16,000 attendees in Vancouver. He actively studies brain development, memory formation, and the neurological impacts of technology on children and young people, and has built an AI clone trained on all his books, research papers, and public television specials to explore these questions firsthand.
That quote applies not just to individual AI tools but to the relationship between AI and early childhood development itself. You can't use AI with young children casually and expect it to be neutral. If you're not actively curating, supervising, and intervening in what the AI teaches, the relationship becomes toxic—the child is absorbing a curriculum you don't control.
The data on related concerns already hints at this risk. The episode also explores how AI affects memory and critical thinking, with research showing that people using ChatGPT often can't recall what they've written minutes later. For children whose brains are still forming memory pathways, this effect could be amplified—and permanent.
Key takeaways
- One-on-one human teaching is the gold standard for early childhood brain development, but remains economically unscalable.
- AI could theoretically scale quality teaching, but the values embedded in AI systems are controlled by whoever designed them—a profound power imbalance.
- Nearly 30% of US parents with children ages 0 to 8 report their kids are already using AI for learning, often without awareness of the cultural and moral curriculum being taught.
- The real concern isn't whether AI can teach, but whether we've solved the problem of whose values, biases, and cultural priorities are being baked into systems that shape young minds.
See also
What is the optimal way to use AI tools to enhance rather than diminish cognitive development and critical thinking?
You have to have a relationship with it or it's going to turn toxic. It's going to hurt you. But if you have a good relationship with it, it can make your life better.
How does passive consumption of AI-generated content affect memory encoding and brain activity compared to interactive engagement?
When people write things with ChatGPT or these AI tools, they don't actually remember, even in some cases minutes later, what they've produced. The brain needs active engagement to encode memory.
What are the neuroscience concerns regarding heavy daily use of AI tools by young people before long-term effects are fully understood?
Before we understand the long-term consequences we can't predict where it's going to end up and it may take 20 years. The real issue is long-term use—we need 30 or 40 years of data to see the full neurological impact, as we saw with calculators.