Originally posted on LinkedIn on June 23, 2026
Whenever my eighty-year-old parents have a computer problem, they call me.
Confusing online banking, a printer that suddenly stops working, a vague iPhone error message — any of these can get them stuck. I don’t mind helping them, but it does make me wonder: when will AI make people like them more self-reliant?
AI already feels very useful in my own life. I use it for coding, office work, research, writing and practical problem-solving. But when I try to get my parents to use it, I’m reminded that AI usability still has a long way to go.
First, choosing the right AI is already confusing. ChatGPT, Claude or Gemini? Free, Plus or Pro? Which model? Which mode? For expert users, this may feel like flexibility. For ordinary users, it looks like a jungle of names, versions and settings.
Then there is the context problem. AI works much better when it understands your situation. But it often doesn’t see what you see, know what you are pointing at, or remember enough background. So the user still has to explain too much.
And then there is reliability. LLMs can still confidently give wrong answers. For important things, you still need to fact-check. That also makes it hard to fully trust AI with your files, accounts or computer.
So yes, AI has become remarkably powerful very quickly.
But for many non-expert users, it still feels like another complicated computer thing they need help with.
Closing that gap may be one of the biggest UX challenges of the next few years.
Back to AI experiments and musings