You do not need prior AI knowledge or a paid account to begin. An AI assistant accepts a request and produces a response; your first skill is deciding which parts of that response deserve trust. Start with synthetic material small enough to check yourself. The examples here are authored teaching examples, not transcripts of model calls or results of simulation.
Day | Lesson and activity | Keep as evidence |
|---|---|---|
1: 20 minutes | Useful AI tasks: classify one task from your discipline. | A scoped request and a reason to use AI. |
2: 20 minutes | Assistants and models: identify what a chat or coding assistant can access. | An allowed-actions statement. |
3: 30 minutes | Your first RTL explanation: read the complete counter and draft an explanation. | A prompt, a corrected explanation and a hand trace. |
4: 30 minutes | Improve and check: correct a claim and hand-trace boundary cases. | A revised answer and independent checks. |
5: 20 minutes | Choose a tool: compare permitted ways to complete this task. | A decision with privacy, time, and cost limits. |
6: 20 minutes | Combine the week's notes into one review packet. | Mark every check as planned, hand-derived, or actually executed. |
7: 15 minutes | Ask a peer to assess the packet, or review it against the exit checks below. | One correction and one next task. |
Produce a short note containing the synthetic counter source, your exact request, the answer you assessed, a corrected explanation, a hand-derived trace, and the tool decision. Label an authored example as synthetic; if you later use an approved assistant, identify its actual response separately. Keep the original mistake visible so a reviewer can follow your correction.
Explain all four ports, reset timing, enable behavior, hold behavior, and wraparound in your own words.
Reject the unsupported frequency claim and name the missing implementation evidence.
Show a reset-with-enable case, a hold case, and a wrap case with explicit initial conditions.
Record who reviewed the note and what remains unverified. A solo review may say 'self-review'; do not invent a reviewer.
Completion means you can distinguish a useful draft from checked engineering evidence. You can finish with the supplied synthetic responses and paper reasoning if no approved assistant is available. RTL, design verification (DV), static timing analysis (STA), and physical design (PD) engineers share this habit even when their eventual tools differ.