I write about AI for a living — what people confessed to me about using ChatGPT surprised me

People feel uneasy about using AI for all sorts of reasons. Environmental concerns come up often. So do worries about cheating at work. But Hass says the strongest emotional reactions tend to happen when people use AI in a specific way.

“It’s the emotionally laden interpersonal messages, or ‘heartfelt messages’,” she explains. “Things like birthday wishes, love letters, wedding vows, and notes of appreciation.”

Sign up for breaking news, reviews, opinion, top tech deals, and more.

Drafting a shopping list with AI is unlikely to keep you up at night. Asking it to help with a work email may feel like common sense. But when a message is meant to signal care, effort and emotional investment, that’s when things get difficult.

“What we find is that it’s not just using AI that creates discomfort,” Hass says. “The guilt comes when messages are sent where the recipient expects genuine personal investment.”

In other words, context matters. “When you use AI to write a birthday card for your best friend, you’re in a situation where honesty, authenticity and effort are core to what the message is supposed to signal. That’s where the negative feelings really kick in.”

What to read next

Hass’s research suggests that the closer the relationship and the more meaningful the occasion, the worse people tend to feel if they rely on AI.

We know people feel bad, but what’s the emotion, specifically?

“The primary emotion we identify is guilt,” Hass says. She explains that the distinction really matters here, because guilt is different from embarrassment or shame. It’s not just about how others might see you, it’s about doing something that feels wrong to you.

“Using GenAI to write a heartfelt message and presenting it as your own creates a sense that you’ve misrepresented yourself to someone you care about,” she explains. “That violation of your own ethical standards is precisely what triggers guilt.”

If you send a love letter, your partner reasonably assumes you sat down and chose those words. That the phrasing reflects your thought and emotional effort. “But then the actual source of those words is an algorithm,” Hass says.

She describes this as a “source-credit discrepancy” — a mismatch between who appears to have authored the message and who actually did. That discrepancy is what makes the act feel so dishonest.

“It’s not just that the message feels abstractly inauthentic,” she adds. “It’s that you’re creating a false impression of authorship in the mind of someone who trusts you.” That’s where the emotional hangover comes from.

When Hass explained this to me, I couldn’t stop thinking about what someone should do if they’re already stewing in this feeling. Should they admit it? Maybe.

Hass says that transparency would likely reduce guilt because it removes the dishonesty at the core of the discomfort. “If the recipient knows the message came from AI, there’s no false impression of authorship, no source-credit discrepancy,” she says. But disclosure doesn’t magically make the situation simple.

Now the dynamic shifts. How does the other person respond? Are they amused? Indifferent? Hurt?

“If they’re fine with it, that acceptance might facilitate a kind of self-forgiveness,” Hass says. “But if they feel let down — that their special occasion didn’t merit your personal effort — that reaction could intensify your guilt, which now comes from hurting someone you care about.”

If you wrote your mum’s birthday card or your wedding vows with AI and aren’t sure what to do, honesty might be the answer. Explaining why you used it, and that you genuinely cared, may help. But this is new emotional territory, and you can’t control how someone else will react.

One obvious option is to draw a hard boundary and avoid using AI for emotionally meaningful communication altogether.

Hass suggests a more practical solution may be to reframe AI’s role in your life. “A more appropriate role for GenAI might be as a thinking partner rather than a ghostwriter,” she says. In my time reporting on AI, that’s broadly the view I hear from people who take a measured approach to AI tools.

“Using it to brainstorm, overcome writer’s block, or refine a draft you’ve already written yourself preserves your genuine voice and investment in the message,” she says. It’s also what she tells her students: “Give it your best, authentic shot first, and then consider whether GenAI can help you sharpen it.”