I Taught My Dad to Use AI. Now He Uses It More Than I Do.

The conventional wisdom about AI adoption is wrong, and the most obvious counterexample lives in my parents' garage in Newbury Park, fixing a sprinkler valve at 7:30 in the morning while listening to ChatGPT explain irrigation timer logic through a Bluetooth speaker.

My dad is 62. He retired from a career in commercial insurance two years ago. He has never written a line of code. He used a flip phone until 2019. The first time I tried to get him to use Google Voice Search in 2017, he held the phone three inches from his mouth and said "please find me a hardware store" in a tone normally reserved for important diplomatic correspondence.

In November of last year I sat him down and showed him ChatGPT for forty-five minutes. He has used it almost every day since. By my rough count, he now opens it more than I do.

I think this is more common than people realize, and I think the cultural narrative around AI adoption has it almost completely backwards.

What he actually does with it

I asked him last weekend to walk me through the last seven days. Here is what he listed, off the top of his head, without any prep.

Tuesday: asked it to translate a technical letter his auto insurance company sent him about subrogation. He told me later that he had received letters like this for thirty-five years and never once fully understood what they were asking for. The translation took eleven seconds.

Wednesday: voice-chatted with it for twenty minutes about how to repair a wobbly chair leg without buying anything new. The chair is fine now.

Thursday: had it draft a 200-word condolence note for an old colleague whose wife had passed. He told me he rewrote about half of it but said the bones gave him something to push against, and he would have stared at a blank page for forty-five minutes otherwise.

Friday: photographed a weird-looking bug on his patio and asked what it was. (Carpenter bee. Harmless. Wood-borer though, so worth keeping an eye on the back deck.)

Saturday: planned a 4-day road trip through Death Valley with my mom. Got a daily itinerary, sunrise viewpoints, restaurant suggestions in Pahrump that turned out to be accurate, and a packing checklist that included things he would never have remembered like extra coolant for the truck.

Sunday: asked it to compare three Medicare supplement plans his neighbor had been bragging about. He had been putting this off for six months because every time he tried to read the brochures, his eyes glazed over by paragraph two. He had a real answer in fifteen minutes.

Monday: wrote a polite-but-firm email to a contractor who was three weeks late on finishing their fence. The contractor showed up Thursday.

That is one week. From a man who, three years ago, asked me how to forward an email.

Why I think this works

The standard explanation for AI adoption among older adults is that it is a substitute for technical skills they never developed. That explanation is technically true and almost completely useless.

The real reason, I think, is that older adults have something younger users do not: a backlog of unanswered questions they have been carrying around, sometimes for decades.

My dad has been receiving incomprehensible insurance and medical paperwork his entire adult life. He has never had a tool he could just hand a confusing letter to and say "what is this person actually asking me." He has had no one to ask, because the obvious people to ask (an actual lawyer, an actual doctor, his actual insurance agent) cost money and time and require an appointment and a phone call. So he has just dealt with each letter as best he could and moved on, accumulating a quiet weight of small confusions across forty years.

ChatGPT cleared that backlog in about three weeks. Not because it answered every question correctly. Because it answered most of them well enough that he could finally stop carrying the ones he had been deferring, and start treating future incoming paperwork as something he could just process, instead of something he had to brace for.

I do not have that backlog. I am 23. The complexity of my paperwork is roughly one tax return and a renter's insurance policy. When I use ChatGPT, I am using it to do new things faster. When my dad uses it, he is using it to finally do old things that have been quietly bothering him for half his life. The marginal value per query is much higher for him than it is for me. By a lot.

The same dynamic, I think, applies to my mom (62, asks it daily for recipe substitutions and to translate text messages from a Spanish-speaking neighbor), my aunt in Camarillo (58, uses it primarily to research her grandkids' homework), and the half-dozen other parents and retirees I have informally surveyed. The pattern is consistent. Older users do not use AI for fewer things than younger users. They use it for fundamentally different things, and the things they use it for tend to have been bothering them for a long, long time.

What teaching him actually looked like

I want to be specific about this, because most of the "how to teach your parents to use AI" content I have seen is condescending and not very practical.

The forty-five minutes I spent with him in November worked because I did three things and skipped four others.

I did: install the ChatGPT app on his phone, log him in with his existing Google account, turn on voice mode, and have him use it three times in front of me. The first time I picked the question (a recipe). The second time I made him pick a question (he asked it to explain why his back hurt after carrying drywall). The third time I made him do voice mode entirely, with no typing.

I did not: explain how large language models work, warn him about hallucinations, tell him to fact-check everything, give him a list of "best prompts," or compare ChatGPT to Claude or Gemini. He does not need any of that yet, and frankly neither do most users.

I told him exactly one thing about accuracy: "If the answer matters and a wrong version of it could cost you money or hurt someone, double-check it with a real source." That is it. He has internalized that rule and uses it correctly. He has, on his own, learned to ask follow-up questions when something seems off, to push back when the model gives a vague answer, and to ignore the parts that read like a brochure.

The only thing I have intervened on since November is voice mode. He tries to talk to it like it is an Alexa command and gets frustrated when it answers in three paragraphs instead of one sentence. I told him to add "give me a one-sentence answer first, and then ask me if I want more detail" to the start of any voice query. He does this now and his frustration has dropped to roughly zero.

The thing nobody is telling parents and retirees

If you are over fifty and reading this and have not started using AI yet, the entire "is this for me" question is the wrong frame.

The right question is: do you have a small pile of recurring small annoyances in your life that are not big enough to pay a professional for, but big enough that they stay at the back of your head? The piles look different for everyone. Insurance paperwork. Medical bills you do not understand. Letters from the IRS. Travel logistics. Home repair questions you would normally ask a younger relative about. Cooking substitutions. Email drafting for difficult conversations. Translating technical jargon in fields you used to work in but have moved on from.

If yes, you have an AI use case right now. You do not need a productivity workflow. You do not need to learn prompt engineering. You do not need to compare paid plans. You need to download one app, open it, and ask it the very next thing on that small pile, in plain English, the way you would ask a smart and patient friend over coffee.

The first three queries are weird. By the tenth, you will have stopped hesitating before you ask it things. By the thirtieth, you will reach for it the way you reach for Google. By the hundredth, you will have a quiet running relationship with a tool that has cleared more low-grade mental clutter out of your life than anything since the original Google search bar.

My dad logged into ChatGPT for the first time on November 14th. He has now used it 1,140 times by his app's count. That is roughly seven times a day, every day, for five and a half months. He is not a power user. He has not signed up for a paid plan. He has not built a single custom GPT. He just uses it the way it is meant to be used, by an actual human with actual questions, and he has gotten more compounded value out of it in five months than I have in the last two years of using it for work.

If you are 60 years old and you have not tried this yet, you will save more time in your first three months of casual use than I have saved in three years of professional use. The math is not close. Open the app, ask it the most boring question on your mind, and start your own count.