Ask AI which skills take little time to learn and offer a high return, and it lists ten. Ask again, and it lists ten more. In Episode 7, this follow-up went on for five rounds, eventually producing a set of criteria: the rules should be reusable and transferable, and ideally used at least once a week.
Regular expressions and browser skills might fit, but what is useful depends on your day-to-day work. Someone who rarely handles text may not need to memorize regex syntax right now; someone who renames files every week can come up with a practice exercise immediately.
Start with a hassle you encountered recently
Look back at the past week and pick an operation you repeated to ask AI about: standardizing filenames by hand, extracting IDs from a column of text, or repeatedly finding the same set of pages in your browser. Explain how often you do it and how you currently do it, and the recommendations will be more relevant to your situation.
Some tasks are best left to a tool. The parts worth learning are usually rules you’ll need again and that affect your ability to judge whether the result is correct. A complex one-off script may not need to become a long-term learning task if you don’t plan to maintain it.
You can also give each candidate skill a very simple test: What real task will you use it for next week? If you can’t say, leave it on the candidate list for now; there’s no need to rush into scheduling a tutorial.
Use a few filenames to find gaps in your understanding
Suppose you want to learn regular expressions. Prepare a set of fictional filenames and practice in plain text or on copies. First write down the date you want to extract, then ask AI to explain the rule. Don’t make batch changes to real files.
After reading the explanation, switch to a different set of inputs and first guess which ones will match. Add a filename with no date, then one with a different date format. Run the rule and compare the results to see where false matches occur.
This kind of practice can distinguish among a few situations: you don’t remember the symbols, you haven’t considered the input format, or you haven’t checked the result at all. Ask AI for a hint about the specific point where you get stuck instead of rereading the whole tutorial.
Try again a few days later to see how much stuck
In their 2013 review of learning techniques, Dunlosky et al. gave practice testing and distributed practice relatively high ratings for applicability. The study discusses learning methods; it does not prove that any particular AI tool is effective. But it can help you plan practice by leaving time to recall and use what you learned again.
A few days later, find another set of filenames and try the task before looking at your original answers. If you can quickly pick up the method again, bring the practice into your everyday work. If you still get stuck, note why and do another round. AI can create exercises and explain things, but this check is still needed to see whether you can do it on your own.
A short note is enough: what input the rule is for, one correct example, and one example that previously produced a false match. When the next task comes up, being able to understand this note and check the output is more helpful than collecting ten more skills.