Power user, but not an expert
Today I asked Claude: How would you rate my AI skills? The answer was nuanced — and a wake-up call.
What I do well:
- Prompt engineering: well above average. I know how to ask an AI the right questions.
- Vibe coding: I build entire platforms with Lovable and Claude Code without writing a single line of code myself.
- Systems thinking: I think in architectures, connect subsystems, plan iteratively.
What I am missing:
- Technical depth: How does a transformer work? What happens during tokenization? Why do models hallucinate?
- Under the hood: I operate the machine, but I don't yet understand its engine.
- Evaluation competence: I can use AI output, but I can't always assess whether it's correct.
This is not a defeat. It's a roadmap. Starting today, for 365 days, I will close this gap. Publicly. Naively. Curiously.
Live from the FAZ AI Conference
This first entry was written on June 23, 2026 at the FAZ AI Conference in Frankfurt. Between talks, during lunch break. Something for everyone — from absolute beginners to silicon nerds. The conference inspired me to keep expanding CompuVerse. And to launch AI Noon right on the spot. At AI speed.
Try it now: Open Claude, ChatGPT or another language model. Ask exactly this question: "How would you rate my AI competence? Be honest." See what happens. Then follow up: "What should I learn first?"