What Your Retired Parents Actually Want From AI (And How to Set It Up in Ten Minutes)

My mother-in-law is seventy-three. She has an iPhone she uses exclusively for phone calls, text messages, and photos of her garden. She does not know what an app does. She knows what a website is. She thinks "the cloud" is a metaphor and is suspicious of it either way.

Last Thanksgiving, someone at the table mentioned that you could ask AI to plan a trip. She looked interested for exactly four seconds, then said, "I wouldn't know where to start," and went back to her mashed potatoes.

That moment — the four seconds of interest followed by the wall of "I don't know how" — is where sixty million Americans over sixty-five live with AI right now. They're not opposed to it. They're not scared of it. They're just standing in front of a door that has no handle.

This article is for their kids. The ones who keep saying "Mom, you should try ChatGPT" without ever sitting down and showing her. The ones who set up an account, texted her the login, and then wondered why she never used it. The ones who are reading this right now thinking, "Yeah, that's my dad."

Here's what your retired parents actually want from AI — and how to set it up so they'll actually use it.

They don't want "AI." They want answers.

The first mistake is talking about AI at all. Your parents don't care about large language models, neural networks, or the fact that Claude is different from ChatGPT. They care about getting an answer to a question without having to parse ten blue links, three ads, and a Reddit thread from 2019.

The pitch is not: "You should use AI."

The pitch is: "There's a thing on your computer where you can type a question in plain English and get a plain English answer. No searching. No clicking. No ads. Want me to show you?"

That's it. That's the whole pitch. If they say yes, you're in. If they say no, you made the offer and you move on. Never push. Pushing creates the opposite of curiosity.

The three things they'll actually use it for

I've watched six people over sixty-five try AI tools for the first time. I've watched what sticks and what doesn't. Three categories stick. Everything else falls off within a week.

1. The question they've been meaning to Google but didn't

Every retired person has a short stack of questions they've been carrying around. Not urgent ones — simmering ones. "What's the difference between Medicare Advantage and Medigap?" "Can I grow tomatoes in a north-facing garden?" "What's the name of that movie with the train — not Murder on the Orient Express, the other one — from the seventies?"

These are perfect AI questions because they're conversational. The person doesn't need to figure out the right search terms. They just ask the way they'd ask a knowledgeable friend, and the AI answers the way a knowledgeable friend would answer.

Set up the tool. Sit with them. Ask them: "What's something you've been meaning to look up?" Watch them type it. Watch the answer come back. Watch their face.

That moment — the moment they realize they can just ask — is the moment it clicks. Everything else builds from there.

2. Letters, cards, and emails they've been putting off

Your father has been meaning to write a thank-you note to his old colleague who sent flowers when your mother was in the hospital. It's been three weeks. He knows what he wants to say but he can't get past the first sentence. He's not blocked because he can't write — he's blocked because it matters and he doesn't want to get it wrong.

AI is spectacularly good at this. Not because it writes better than your father — it doesn't — but because it gives him a draft to react to. "Here's a version. What would you change?" is easier than a blank page, every time.

The prompt that works: "Help me write a thank-you note to my friend Bill who sent flowers when my wife was sick. I want it to sound like me — not fancy, just sincere. Two or three sentences."

The key phrase is "sound like me." Without it, the AI will produce something generic and your father will say, "That doesn't sound like something I'd write," and close the tab forever. With it, the AI stays close to simple, direct language — which is usually what the person actually sounds like.

This extends to: birthday cards for grandchildren, emails to doctors' offices, letters to insurance companies, condolence notes, complaint letters, and the annual holiday update that takes three weeks to draft.

3. Planning things that used to require a travel agent

Your mother wants to visit her sister in Savannah. She wants to fly, not drive. She wants a hotel near the historic district but not in the middle of the noise. She wants to know which restaurants have outdoor seating and aren't too expensive. She wants to know if the garden tours are still running in October.

This is one question. But on Google, it's seventeen searches, four tabs, and a headache. On an AI, it's a conversation: "I'm visiting my sister in Savannah for four days in October. I'm seventy-three. I want to fly from Philadelphia. I want a quiet hotel near the historic district, walking-friendly restaurants, and garden tours. What do you suggest?"

The AI will give her a rough itinerary she can react to. "That hotel looks good but move dinner to somewhere quieter." "Are there any good bookstores near that area?" Each follow-up makes the plan more hers.

This is the use case where I've seen the most delight. Planning a trip used to mean calling a travel agent or spending a whole Sunday afternoon on the computer. Now it takes twenty minutes and feels like talking to a concierge who has infinite patience.

How to set it up so they'll actually use it

The setup matters more than the tool. Get this wrong and they'll never open it again.

Step 1: Choose one tool, not three. Claude, ChatGPT, Gemini — pick one. Create the account yourself using their email. Log them in on their computer or phone. Don't explain the alternatives. Choosing is a barrier.

Step 2: Put it where they can find it. A bookmark on their browser's bookmark bar, labeled with something they'll recognize. Not "Claude" — they don't know what Claude is. Label it "Ask a Question" or "My AI Helper." On a phone, put the app icon on the home screen, first page, near the phone and messages icons.

Step 3: Sit with them for the first three questions. Not over FaceTime. In person, if you can. Side by side, so they're driving and you're just there if they get stuck. Let them type. Let them see the response. Let them ask the follow-up. Don't take the keyboard.

Step 4: Write down three prompts on a physical piece of paper. Yes, paper. Tape it next to their computer. Three prompts they can type in verbatim when they want to try it on their own:

• "I have a question about [topic]. Explain it simply."
• "Help me write a short note to [person] about [reason]. Keep it simple and sincere."
• "I'm planning a trip to [place] for [number] days. I'm [age] and I want [preferences]. What do you suggest?"

These are training wheels. They'll outgrow them in a week. But the first week is the one that matters.

Step 5: Check in after three days. Not to teach — to ask. "Did you try it? What did you ask?" If they did, celebrate whatever they tried. If they didn't, offer to sit with them again. No guilt.

The mistakes their kids make

Mistake 1: Showing off. You open the AI and ask it to write a sonnet or generate code or explain quantum computing. Your parent watches politely and thinks: "This is not for me." Show them something that solves a problem they already have. The sonnet can wait.

Mistake 2: Sending a link. "Here, I set up an account, here's the link." Links die. Bookmarks get lost. If you're not there to walk them through the first session, the link is just another tab they'll close.

Mistake 3: Explaining how it works. They don't need to know about training data, parameters, or hallucinations. They need to know: "Type your question. Read the answer. If the answer doesn't make sense, ask a follow-up." That's the entire user manual.

Mistake 4: Giving up after one try. Your father tried it, got a weird response, and said "this thing doesn't work." That's not failure — that's a bad prompt. Sit with him, rephrase the question together, and show him that the second try usually works. The tool improves when the question improves. So does every conversation with a human.

What they're afraid of (and what to say about it)

"Will it steal my information?" Honest answer: "The things you type are used to generate a response. I wouldn't put your Social Security number or bank password in there, the same way you wouldn't give them to a stranger on the phone. But asking about garden tomatoes or writing a thank-you note? That's fine."

"What if it gives me wrong information?" Honest answer: "It sometimes does, the same way a person sometimes does. For anything important — medical decisions, legal questions, financial choices — use it to understand the topic, then verify with a real professional. For trip planning and letter writing, it's reliable."

"I'll break it." Honest answer: "You literally cannot break it. There is no wrong button. Type anything. The worst that happens is it gives you a weird answer, and then you just ask again."

The thing nobody talks about

There's a loneliness dimension to this that most articles about "AI for seniors" skip, because it's uncomfortable.

Some retired people live alone. Their kids call on Sundays. Their friends are on different schedules, or have moved, or have died. The house is quiet. They have thoughts and nobody to share them with — not big existential thoughts, just the Tuesday afternoon kind. "I wonder why cardinals are red." "What was the name of the place where Dad and I ate in Rome in 1987?" "Can someone explain what my grandson means when he says something is 'mid'?"

AI handles these questions without judgment, without impatience, and without the subtext of "you're bothering me." It isn't a substitute for human connection — nothing is. But it is a thing that answers when you ask, and for someone who spends a lot of hours with the television, that's not nothing.

I'm not going to oversell this. Your mother doesn't need AI. She lived seventy-three years without it and she's fine. But if there's a version of her life where she types a question at 2 PM on a Tuesday and gets an answer that makes her smile or saves her an afternoon of frustration — that's a better Tuesday. And Tuesdays add up.

Set it up next time you visit. Sit with her. Let her ask the first question.

You might be surprised what she wants to know.