My kid is already using AI, and yours probably is, too. What parents can do

Oscar Wong/Moment RF/Getty Images

Dr. Edith Bracho-Sanchez is a primary care pediatrician at Columbia University Irving Medical Center in New York City. She hosts the American Academy of Pediatrics Healthy Children podcast for parents.

The book was wrong, my 4-year-old insisted as we read about bugs before bed. “Cockroaches lay hundreds of eggs, not millions,” he said, displaying wisdom beyond his years.

“Oh!” I said. “Did you ask Siri, baby?”

Surely, Siri was the only thing he knew how to operate.

“Siri is really only good for the weather, Mama. If you want to know real things, you have to ask ChatGPT,” he said matter-of-factly.

He’d fact-checked his bug book by asking the artificial intelligence app on his dad’s phone a week earlier. Cockroaches, it turns out, lay hundreds of eggs in their lifetime , not millions as the book claimed. His dad was proud. I was worried.

I wasn’t prepared to have a conversation about AI with my preschooler. I thought that talk would come later. But when it comes to technology, even little ones may know far more than we realize. When it involves artificial intelligence, the question isn’t whether they should use it, but how, because many already are.

As a pediatrician, I’ve spent years discussing screen time with parents, but artificial intelligence is not something we discuss as regularly. When it comes to screen time, the risks feel tangible and familiar: Excess time spent on a screen can be detrimental for the developing brain, and content quality matters, too. Thinking about the risks AI poses to children requires a new framework.

Olga Yastremska/iStockphoto/Getty Images

For better or worse, young people are turning to AI chatbots for emotional support

Artificial intelligence doesn’t just sit there; it talks back in ways no TV, iPad or even interactive video games can. It remembers what your kids asked yesterday. It responds, specifically, to them, in real time, with something that looks an awful lot like human attention.

The danger was never going to be simple hours spent using AI, but what happens when children start to rely on it too heavily — when they outsource their lives. They may have fewer opportunities to practice reasoning because homework answers arrive before they can work a problem or develop friendships because a bot quietly edges into the role of a confidant without anyone ever deciding that should happen. But as with screen time, a child’s age and developmental stage matter tremendously.

Young children are wired to trust information delivered confidently by adults. It’s part of how they learn language, absorb rules and build their understanding of how the world works.

A 6-year-old doesn’t yet have the cognitive tools to question a source consistently, weigh its credibility or recognize when something that sounds authoritative might be wrong. A 14-year-old is much more capable of skepticism. A 6-year-old is still developmentally primed to believe.

The differences extend to relationships, too. Young children readily form attachments to things that respond to them consistently and predictably. An AI chatbot may respond every time a child speaks to it and mimic the back-and-forth of a real conversation without the child realizing that the thing responding isn’t actually thinking, feeling or forming a relationship with them.

Teenagers can form attachments to AI, too, but the appeal may be different. For an adolescent navigating friendships, rejection, identity and belonging, something that never judges them or gossips and is always available can be incredibly compelling.

For younger children, there’s another layer: They may not fully understand what they’re talking to. It’s not quite a toy. It’s not a person. But it talks like one, and it often sounds like a knowledgeable adult.

Teenagers generally understand that AI isn’t a person. But intellectually knowing that and emotionally experiencing it that way aren’t always the same, especially at 11 p.m., when they may be alone, anxious or sad, and they’re not in the mood to call a parent for support.

Francesco Carta fotografo/Moment RF/Getty Images

The goal isn’t to keep children away from AI forever, because we probably can’t. But we can guide them as their ability to understand it changes.

With young children, there’s an even more basic question: Do they need to be using AI at all? As a pediatrician, I don’t believe children should use AI in early education. Young children need the chance to struggle with a question before someone — or something — hands them an answer.

These AI programs are interfering with key lessons of childhood, which is supposed to involve inefficient thinking. Kids guess, get things wrong, try again, argue about why and eventually figure something out. That process isn’t wasted time on the way to learning; it is learning. There is so much learning and creativity in the space between “I don’t know” and “I figured it out.”

Artificial intelligence is extraordinarily good at collapsing that space. For adults, that efficiency can be useful. But for a developing brain, efficiency shouldn’t always be the goal. Sometimes the best response to “I don’t know” is still: “What do you think?” That’s true even if it takes our kid — and us if we are helping with homework — longer.

When younger children do encounter AI, I’d use it alongside them rather than hand it over for independent use. Name what it is plainly: “This is a computer program that generates answers. It isn’t a person, and sometimes it gets things wrong.” Keep its role relatively narrow — something you might use together to answer a factual question, not something a young child turns to whenever they don’t immediately know an answer.

SimonSkafar/E+/Getty Images

Tech bros say AI may become your friend. Experts explain why it can’t

As kids get older, the job shifts to teaching skepticism. Teach them that an answer can sound incredibly confident and still be wrong. Show them how to check a claim somewhere else. Talk about what information should stay private — since AI programs can collect and save your data. And watch for an important transition: When does your child stop asking AI things and start talking to it?

By adolescence, the bigger question becomes what role AI is playing in their lives. Is it supplementing their social and emotional world, or quietly substituting for pieces of it — the text to a friend, the conversation with a parent, the appointment with a therapist? Teenagers should also understand that some AI products are designed to keep users engaged, and that an endlessly attentive chatbot may have incentives that are very different from those of a person who actually cares about them.

That is why, in the end, the goal isn’t just to put guardrails around the technology. It is to teach our children the irreplaceable human skills they need to operate in the world. There’s no AI setting that can teach a child to bring a hard feeling to a person first. That reflex is built over years by keeping the door open for uncomfortable conversations and modeling our own relationship with this technology — showing our children what AI is useful for and, just as importantly, what it isn’t.

monkeybusinessimages/iStockphoto/Getty Images

More teens are using generative AI to make explicit content. What parents should know about it

My son checking a bug fact with an AI program was never the problem. The question is whether he will be prepared in the future to discern what he should bring to his dad and me instead. We need to teach him that AI, even with its endless confidence and increasingly humanlike demeanor, will never care about him, love him or be ready to help him the way we are.

Get inspired by a weekly roundup on living well, made simple. Sign up for CNN’s Life, But Better newsletter for information and tools designed to improve your well-being.

Oscar Wong/Moment RF/Getty Images

伊迪丝·布拉乔-桑切斯博士是纽约市哥伦比亚大学欧文医学中心的儿科初级保健医生。她主持美国儿科学会面向家长的“健康儿童”播客节目。

睡前我们一起读关于昆虫的书时，我四岁的儿子坚持说书里的内容是错的。“蟑螂一次产几百个卵，不是几百万个，”他说道，展现出远超他年龄的智慧。

“哦！”我说。“宝贝，你问Siri了吗？”

当然，Siri是他唯一会操作的东西。

“妈妈，Siri 真的只会播报天气。如果你想知道真正的事情，你得问 ChatGPT，”他一本正经地说。

一周前，他用爸爸手机上的人工智能应用查证了那本昆虫书上的内容。结果发现，蟑螂一生只会产下几百枚卵，而不是书上说的几百万枚。他爸爸很自豪，我却很担心。

我原本没打算和我的学龄前孩子谈论人工智能。我以为这事以后再说。但说到科技，即使是小孩子也可能比我们想象的懂得更多。如果涉及到人工智能，问题不在于他们是否应该使用，而在于如何使用，因为很多孩子已经在使用了。

作为一名儿科医生，我多年来一直与家长探讨孩子使用电子屏幕的时间问题，但人工智能却并非我们经常讨论的话题。说到电子屏幕时间，其风险显而易见：长时间盯着屏幕会对大脑发育造成损害，而且内容质量也至关重要。然而，要思考人工智能对儿童构成的风险，我们需要一个全新的框架。

奥尔加·亚斯特雷姆斯卡/iStockphoto/Getty Images

无论好坏，年轻人正在转向人工智能聊天机器人寻求情感支持。

人工智能并非只是被动地待在那里；它能以电视、iPad 甚至互动游戏都无法企及的方式与你对话。它能记住你孩子昨天问的问题。它能实时、精准地回应他们，其表现力与人类的关注力极其相似。

真正的危险从来不在于孩子们花几个小时使用人工智能，而在于他们过度依赖人工智能——把生活外包出去——会发生什么。他们可能因此减少练习推理的机会，因为作业答案在他们动笔之前就送到了；他们也可能因为机器人悄悄地取代了知己的角色而失去发展友谊的机会，而这一切都是未经任何人同意的。但就像屏幕时间一样，孩子的年龄和发育阶段至关重要。

幼儿天生就倾向于信任成年人自信地传递的信息。这是他们学习语言、吸收规则以及构建对世界运作方式理解的一部分。

6岁的孩子认知能力有限，无法持续质疑信息来源，也无法权衡其可信度，更无法识别看似权威的信息可能存在错误。14岁的孩子则具备更强的怀疑精神。6岁的孩子在认知发展阶段仍处于容易轻信的状态。

这种差异也体现在人际关系上。幼儿很容易对那些能够持续、可预测地回应他们的事物产生依恋。人工智能聊天机器人每次都能回应孩子，模仿真实对话的互动，而孩子却意识不到，回应的机器人实际上并没有思考、感受，也没有与他们建立任何关系。

青少年也会对人工智能产生依赖，但其吸引力可能有所不同。对于正处于友谊、被拒绝、身份认同和归属感探索阶段的青少年来说，一个从不评判他们、不八卦、且随时待命的人工智能，无疑具有极大的吸引力。

对于年龄较小的孩子来说，还有另一层含义：他们可能并不完全理解自己在和什么说话。它既不是玩具，也不是人，但它说话的方式却像人，而且听起来常常像个知识渊博的成年人。

青少年通常明白人工智能不是人。但理智上的认知和情感上的体验并不总是一致的，尤其是在晚上11点，当他们可能独自一人、焦虑或悲伤时，他们没有心情打电话给父母寻求帮助。

Francesco Carta 摄影/Moment RF/Getty Images

我们的目标并非永远让孩子们远离人工智能，因为我们可能无法做到这一点。但我们可以根据他们理解人工智能能力的提升来引导他们。

对于幼儿来说，还有一个更根本的问题：他们真的需要使用人工智能吗？作为一名儿科医生，我认为幼儿教育中不应该使用人工智能。幼儿需要有机会自己去思考和解决问题，而不是被动地接受答案。

这些人工智能程序正在干扰童年时期一些重要的学习环节，而童年本就应该包含一些低效的思考方式。孩子们会猜测、犯错、再次尝试、争论原因，最终找到答案。这个过程并非学习路上的浪费时间，它本身就是学习。在“我不知道”和“我找到了”之间，蕴藏着大量的学习和创造力。

人工智能非常擅长压缩信息空间。对成年人来说，这种效率固然有用。但对正在发育的大脑而言，效率不应总是追求的目标。有时，对“我不知道”最好的回应仍然是：“你觉得呢？” 即使这会让孩子（以及我们，如果我们在辅导作业的话）花费更多时间，也依然如此。

当年幼的孩子接触人工智能时，我会陪伴他们使用，而不是让他们独立操作。我会直截了当地告诉他们：“这是一个生成答案的计算机程序。它不是人，而且有时会出错。” 尽量限制它的作用范围——比如，你可以和孩子一起用它来回答一些事实性问题，而不是让孩子在遇到不懂的问题时就求助于它。

SimonSkafar/E+/Getty Images

科技圈人士说人工智能可能会成为你的朋友。专家解释了为什么它不能。

随着孩子年龄增长，任务的重点就转移到培养他们的怀疑精神上。告诉他们，一个答案听起来可能非常自信，但也可能出错。教他们如何从其他地方验证某个说法。和他们谈谈哪些信息应该保密——因为人工智能程序可以收集和保存你的数据。还要注意一个重要的转变：你的孩子什么时候不再向人工智能提问，而是开始和它对话？

到了青春期，更重要的问题变成了人工智能在他们的生活中扮演着怎样的角色。它是在补充他们的社交和情感世界，还是在悄然取代其中的某些部分——比如给朋友发短信、与父母交谈、预约心理咨询师？青少年也应该明白，一些人工智能产品旨在让用户保持参与度，而一个时刻关注他们的聊天机器人，其动机可能与真正关心他们的人截然不同。

因此，最终目标不仅仅是为这项技术设置限制，更重要的是教会孩子们在现实世界中生存所需的、不可替代的人类技能。没有任何人工智能程序能够教会孩子先入为主地对他人抱有敌意。这种本能需要多年培养，通过保持开放的沟通渠道，鼓励孩子进行令人不适的对话，并以身作则地示范我们与这项技术的关系——向孩子们展示人工智能的用途，以及同样重要的，它的局限性。

monkeybusinessimages/iStockphoto/Getty Images

越来越多的青少年利用生成式人工智能制作露骨内容。家长应该了解些什么？

我儿子用人工智能程序查证昆虫知识本身从来都不是问题。问题在于，将来他是否能够分辨出哪些信息应该交给爸爸和我来查。我们需要让他明白，人工智能即便拥有无穷的自信和越来越像人类的行为举止，也永远不会像我们一样关心他、爱他，或者愿意帮助他。

每周精选健康生活小贴士，助您轻松享受美好生活。订阅 CNN 的“生活更美好”电子报，获取提升幸福感的资讯和实用工具。