What begins as help with homework can quickly develop into something feigning a personal relationship. But, as Turkle warns, chatbots weaken the skills needed to relate to other people.

Consider listening. “We don’t have to listen” to a chatbot when we talk to it, she points out. “We just say our thing and it tells us we’re great … we become de-skilled as human listeners and as givers.” Chatbot conversations are transactional. Ask and receive. But in the real world, relationships require us to give, not only receive. 

Then there’s the capacity for vulnerability. “It turns out that there’s no intimacy without vulnerability,” says Turkle. “And what chatbots offer is an alternative where you can have something that looks like friendship, but without the vulnerability … you don’t have a real connection.” 

Chatbots have all the words in the world, but they haven’t lived the life to back them up. Real relationships need more than just words. They require listening, compromise and a willingness to put someone else’s needs before our own.

Then there’s the impact chatbots have on a child’s thinking skills.

Turkle echoes a 2025 MIT study that explored the impact of AI on the writing process. “When students write a paper with a chatbot, they don’t remember what’s in their paper,” she explains. Even if they dress it up a little, they won’t remember it well, because they delegated the thinking that produced it.

Over time, this can lead to what the study’s authors call cognitive debt, where reliance on AI models “replaces the effortful cognitive processes required for independent thinking.” Possible long-term effects? Diminished critical inquiry, increased vulnerability to manipulation and decreased creativity.

At best, students who use chatbots are gaining only superficial comprehension of their subjects. When pupils are assigned a difficult text, it’s tempting to plug it into an LLM tool and get a summary, podcast or infographic.

“That’s not reading,” says Turkle. When a text has big words, new concepts or an antiquated writing style, it takes time and effort to gain comprehension. Letting ChatGPT or Gemini take the wheel is a missed opportunity. 

So what can parents do?

Turkle says moms and dads are the first line of defense. “I think that this is going to be a parents’ movement first,” she says. “If you see that the school is moving heavily to an AI curriculum … just say, ‘No, I don’t want my child in a school that has this.’ ” Unite with other parents who want to protect their kids from these risks.  

At home, a family policy on chatbots should start as early as possible. “The conversation about how this is not a part of a social life has to start when they’re very young,” Turkle advises. She recommends clear language: “In our family, we don’t use it because it’s dangerous. We don’t smoke in our family, we don’t take drugs in our family and we don’t do this.”

If you do allow your student to use a chatbot, train them to treat it “like a glorified Google.” Supervise the interaction, keep it brief and instruct the chatbot not to use personal pronouns.

Turkle is careful to point out that her concern is artificial intimacy, not AI entirely. 

“There are so many ways this technology can help us,” she says. “I’m not saying that there won’t be interesting educational uses for generative AI, but it’s not going to be from chatting with chatbots.” 

Andrew McDiarmid is a senior fellow at the Discovery Institute. He writes about technology on his Substack The Human Adventure.