Is an AI-Written Birthday Card Still From You?

Ask a 10-year-old whether a friend should use AI to write a birthday card and you won’t get a simple yes or no. You’ll get a question back: how much AI? When University of Washington researchers posed exactly that scenario to kids aged 7 to 13, the children “started to have really nuanced takes,” lead author Michele Newman told UW News. Some asked whether the friend used it “to write the whole card, or just to help.” They wondered whether something is lost when AI, rather than a friend, writes the card.

So is an AI-written birthday card still from you? The kids’ instinct, that it depends on what the AI did, holds up well against what researchers have found about trust, authorship and disclosure. Here’s how to turn it into a family rule.

Key Takeaways

• In a UW study presented at CHI 2024, kids 7 to 13 gave nuanced answers about AI-written birthday cards, asking how much the AI did. One 11-year-old said AI-written books would “dismantle” the joy of reading for him.
• In 13 experiments, people who disclosed AI use were trusted less than those who didn’t, but being exposed by someone else hurt trust more (Schilke & Reimann, 2025).
• People tend not to feel they own AI-generated text, yet often don’t say AI wrote it (Draxler et al., 2024). More personal input raised ownership.
• A personal message’s value comes partly from the attention it shows. The more the AI supplies, the less of that signal remains.
• A workable family rule: the feeling and key words are yours; AI can help with spelling or ideas; say so when it helped.

What Kids Said About an AI-Written Birthday Card

The finding comes from a co-design study by the University of Washington’s KidsTeam, presented May 14, 2024, at the ACM CHI Conference on Human Factors in Computing Systems. Twelve Seattle-area children aged 7 to 13 worked with researchers using OpenAI’s ChatGPT and DALL-E and Google’s music generator Magenta. The paper’s title is a child’s quote: “I want it to talk like Darth Vader.”

Newman described the ethics conversations in an interview with UW News. When asked about cheating, the kids “tended to reiterate things they’ve heard,” like “I shouldn’t use it to cheat.” But the birthday card question was different. “Some started asking how much the friend is using it. Is it to write the whole card, or just to help? Every kid starts to have different ideas.”

Two more details from the interview stand out. One 11-year-old, asked how he’d feel if his favorite book series had been written by AI, said it would “dismantle” the joy of reading. And Newman framed the card question as one the kids were taking seriously: “They’re asking whether they lose some authenticity when AI rather than a friend writes a birthday card.”

A caveat: this is a small qualitative study, 12 children in one region, reported here through a university news interview. It tells us how kids reason, not what most kids believe.

Why the Same Words Mean Less When AI Writes Them

Communication researchers have a term for this: AI-mediated communication. Jeffrey Hancock, Mor Naaman and Karen Levy defined it in the Journal of Computer-Mediated Communication (2020) as interpersonal communication “in which an intelligent agent operates on behalf of a communicator by modifying, augmenting, or generating messages.” A birthday card written by a chatbot is a textbook case.

Here’s the mechanism. A handwritten card carries two messages. The first is the text: “Happy birthday, you’re the best.” The second is the signal underneath: I spent time thinking about you, specifically. The second message is what makes a card from a friend different from a card from a bank. When AI generates the text, the first message can get better while the second gets weaker, because the effort and attention it was supposed to prove didn’t happen.

That’s why kids zero in on “how much.” Fixing spelling barely touches the signal. Generating the whole card replaces it.

People do react to suspected AI

An early experiment by Maurice Jakesch and colleagues at CHI 2019 tested whether people trust Airbnb host profiles less if they think AI wrote them. They found what they called the “Replicant Effect”: when participants saw a mix of AI- and human-written profiles, they mistrusted the hosts whose profiles were labeled as, or suspected to be, written by AI.

Disclosure costs something, but hiding costs more

Oliver Schilke and Martin Reimann reported 13 experiments in Organizational Behavior and Human Decision Processes (2025). Across tasks from communications to artistry, “actors who disclose their AI usage are trusted less than those who do not.” That’s the uncomfortable part. The important part for kids is the comparison: the disclosure penalty was “comparatively weaker than the effect of third-party exposure.” Telling someone you used AI costs a little trust. Having them find out from someone else costs more.

People don’t feel like authors, and don’t say so

In two studies, Fiona Draxler and colleagues (ACM Transactions on Computer-Human Interaction, 2024) found that people “do not consider themselves the owners and authors of AI-generated text but refrain from publicly declaring AI authorship.” They called it the AI Ghostwriter Effect. Higher personal influence on the text increased the sense of ownership. For a card, that suggests a clear path: the more of the words are your child’s, the more the card is honestly from them. We look at why kids hide AI help in why kids hide their AI use.

All three of these studies are with adults. None tested children giving cards to friends. The kids’ own reasoning in the UW study, though, lines up with them.

AI in a Card: A Family Decision Table

What AI does	How much of “you” is in it	Does it feel like it’s from you?	Say AI helped?
Fixes spelling or grammar	Nearly all	Yes	Not needed, same as spellcheck
Suggests a rhyme for a line your child wrote	Most	Yes	Optional, nice to mention
Gives three ideas; your child picks one and writes it in their words	Most	Mostly	A good idea
Writes a draft; your child edits a few words	Little	Weakly	Yes
Writes the whole message; your child copies it out	Very little	Not really	Yes, or don’t send it as personal
Makes the picture on the front from your child’s description	Idea is theirs, art isn’t	Partly	Yes (“I asked AI to draw my idea”)

How Families Can Decide

Start with the kid’s own question

Ask your child the UW question directly: “If a friend used AI to write you a card, would it still feel like it’s from them?” Then follow up with “What if they only used it for spelling? What if it wrote the whole thing?” Kids usually build the scale in the table above on their own. A rule they reasoned their way into sticks better than one you hand them.

Set the “feeling and key words” rule

A simple family standard: the feeling and the important words come from you. AI can help with spelling, a rhyme, or an idea when you’re stuck, but the message itself should sound like you and say something only you would say. A detail from a shared memory is the best test. AI can’t supply “remember when we got lost at the zoo?”

Make disclosure normal and light

If AI helped meaningfully, add a line: “I had help from AI with the poem, but the jokes are mine.” The Schilke and Reimann results suggest telling costs less trust than being found out. Kids can see that tradeoff clearly when it’s framed as a choice. For school writing, our guide to citing AI in homework applies the same idea.

Keep handwriting in the loop for young kids

For ages 5 to 8, the simplest version is a handwritten card, even three words and a drawing. There’s no AI question to answer, and it’s the clearest signal of attention a young child can give. Save AI-assisted messages for when kids are old enough to decide what to disclose.

What not to do

Don’t shame a child who used AI for a card. Their motive is usually good: they wanted it to be nicer than they could make it. Talk about what the recipient will value instead. Most kids, asked whether Grandma would rather have a perfect poem or a messy one that mentions their fishing trip, pick the messy one.

What to Watch For Over the Next 3 Months

• Week 4: Your child can explain your family rule in their own words and apply it to a new case (a thank-you note, a group-chat message).
• Month 2 red flags: Messages to friends and family increasingly sound generic or adult. Your child hides AI use in personal messages, or gets defensive when asked.
• Month 3 self-check: Your child sends a card or message with a specific personal detail only they could have written, and mentions AI help when there was some.

Frequently Asked Questions

Is it wrong for my kid to use AI to write a birthday card?

Not wrong, but it changes what the card means. A personal card signals time and attention. If AI writes the whole thing, that signal weakens. Using AI for spelling or an idea, while keeping the feeling and key words your child’s own, keeps the card genuinely from them. Saying when AI helped keeps it honest.

What did kids in the UW study think about AI birthday cards?

They didn’t give a single answer. When asked whether a friend should use AI to write a card for them, kids aged 7 to 13 asked how much AI was used: the whole card or just help. Researchers described their views as “really nuanced.” Some wondered whether a card loses authenticity when AI writes it.

Should my child tell people when AI helped write a message?

Research with adults suggests yes, on balance. In 13 experiments, people who disclosed AI use were trusted somewhat less, but being exposed by a third party hurt trust more. For personal messages, a light line like “AI helped with the rhyme” is usually enough. Kids tend to understand the tradeoff when it’s explained.

How can AI help without making a card feel fake?

Let it handle the parts that don’t carry the feeling: spelling, a rhyme, a layout idea. Keep the parts that do carry it: a shared memory, an inside joke, why this person matters. People feel more ownership of AI-assisted writing the more they influence it, and recipients notice personal details that no AI could know.

---

About the author

Ricky Flores is the founder of HiWave Makers and an electrical engineer with 15+ years of experience building consumer technology at Apple, Samsung, and Texas Instruments. He writes about how kids learn to build, think, and create in a tech-saturated world. Read more at hiwavemakers.com.

---

Sources

1. University of Washington News. (2024, May 29). “Q&A: How AI affects kids’ creativity” (interview with Michele Newman on the CHI 2024 paper “I want it to talk like Darth Vader”). https://www.washington.edu/news/2024/05/29/ai-kids-creativity-chatgpt/
2. Schilke, O., & Reimann, M. (2025). “The transparency dilemma: How AI disclosure erodes trust.” Organizational Behavior and Human Decision Processes. https://doi.org/10.1016/j.obhdp.2025.104405
3. Draxler, F., Werner, A., Lehmann, F., Hoppe, M., Schmidt, A., Buschek, D., & Welsch, R. (2024). “The AI Ghostwriter Effect: When users do not perceive ownership of AI-generated text but self-declare as authors.” ACM Transactions on Computer-Human Interaction. https://doi.org/10.1145/3637875
4. Hancock, J. T., Naaman, M., & Levy, K. (2020). “AI-mediated communication: Definition, research agenda, and ethical considerations.” Journal of Computer-Mediated Communication, 25. https://doi.org/10.1093/jcmc/zmz022
5. Jakesch, M., French, M., Ma, X., Hancock, J. T., & Naaman, M. (2019). “AI-mediated communication: How the perception that profile text was written by AI affects trustworthiness.” CHI Conference on Human Factors in Computing Systems. https://doi.org/10.1145/3290605.3300469