Black Teens Use AI Chatbots Three Times as Often as White Teens. Here's What's at Stake.
Contributing writer Anya Armentrout reports on new Pew Research Center data showing Black teens use AI chatbots for schoolwork, news and emotional support at significantly higher rates than white teens. Psychologist Tascha Just and cognitive psychologist Brian Stone weigh in on cheating, cognitive dependence, cultural bias in AI-generated writing, and the risks of turning to chatbots for emotional support, especially as the technology widens rather than narrows educational gaps.
A recent study by the Pew Research Center found that 18% of Black teens ages 13-17 use artificial intelligence (AI) chatbots to help with all or most of their schoolwork, triple the 6% of white teens. And 38% of Black teens say AI chatbots have been useful or extremely useful for completing their schoolwork, nearly double the 22% of white teens.
AI and large language models have entered classrooms and students’ lives rapidly, but some psychologists warn that AI-supported schoolwork doesn’t always lead to real learning.
“[AI] can be a support for creativity and to access education, it can be a tutor, it can help organize and help future career readiness, but it can also be a privacy risk and a cheating trap and a surveillance tool,” said Tascha Just, President of the Minnesota School Psychologists Association.
AI can be used to help complete schoolwork in many ways, and the line between what is and isn’t cheating isn’t always clear. Some students use chatbots as research platforms, others to explain concepts, and others for editing or co-writing answers. Policies around what is or isn’t allowed aren’t consistent.
“AI is sort of asymmetrically being introduced in some places in education and not other places, some schools, not other schools, by some teachers and not other teachers,” said cognitive psychologist Brian Stone.
Stone’s work has shown that more than 60% of college students admit to using AI to cheat. A separate Pew study from 2024 found that one in four teachers think AI will do “more harm than good” in education.
“Most teachers I speak with are really nervous about cheating,” Just said. “Critical thinking and cognitive dependence are another issue. It can help critical thinking if you use the right prompts to help you learn, but then it can also just spit out an answer.”
When students become cognitively dependent on technology, they offload mental tasks so often that they aren’t able to complete them independently. Stone likens the brain to a muscle.
“If you’re not careful, it’s like if you go to the gym every day, but have a robot lift the weights for you,” Stone said. “You can brag about your new personal record bench press this week or the A you got on a paper, but if it’s a robot that did it, then your own muscles are just getting weaker and weaker. In this case, your brain.”
“I think students just need to be doing hard things,” he added. “If AI facilitates that, great, they will learn well or even learn better. But if AI helps them avoid the effort, I think that’s where the risk is.”
Just pushes back on those concerns. “I think there’s a lot of fear mongering with regard to that,” she said. “When you find a helpful tool, you use it. I depend on my glasses every single day, and that used to be something controversial.”
She often works with special needs students for whom AI opens new possibilities by using deepfake technology with shy students, translating drawings into words for nonverbal students, and acting as a “neurotypical translator” for students on the spectrum.
Stone warns that even co-writing with AI, where students generate work and then edit and integrate it with their own writing, produces shallower learning. He’s also observed what he calls “the Google effect:” when someone knows information is readily available, their brain is less likely to retain it. The same applies to AI.
With Black youth using AI for schoolwork three times as often as white youth, the technology threatens to widen educational outcome gaps. Particularly when some districts lack resources to teach AI literacy or maintain tech-free spaces.
“I think it’s about the support to use it well,” Stone said. “It’s entirely possible rich school districts are going to be better at putting the technology away and being able to dedicate the time to teach students hands-on.”
There’s also a cultural dimension to AI-produced content. Large language models are trained on existing data, which may overrepresent certain voices and flatten others.
“ChatGPT will push you toward a very standardized voice,” Stone said. “The data comes from our culture, from our books, which might overrepresent certain voices, and so the AI is also going to overrepresent those voices.”
That bias extends to enforcement. AI-detection tools are more likely to flag writing by non-native English speakers as artificially generated, raising equity concerns about who gets accused of cheating.
“The emotional impact of being accused of cheating when you’re not is so significant and can be really traumatic for kids,” Just said.
Black youth also use AI chatbots more for non-academic purposes. 29% use them to get news (compared to 15% of white youth) and 21% use them for emotional support or advice (compared to 8% of white youth). Just finds that last statistic especially concerning.
“These are bots that have not been trained by clinically-trained professionals,” she said, “and it is nerve-wracking to think about what non-professionals might recommend.”
Pew found that teens overall are more hopeful than not about AI, actively weighing the pros and cons. But with Black youth engaging the technology at disproportionate rates, for both schoolwork and emotional guidance, the stakes of getting it right are especially high.
Anya Armentrout is a freelance journalist, a student at Macalester College, and a contributing writer for the Minnesota Spokesman-Recorder. She welcomes reader responses at aarmentr@macalester.edu.