TL;DR: A simple log of when and how you use AI reveals patterns you can’t see in real-time. This awareness is what separates people who control their AI use from people whose AI use controls them.

The Short Version

You probably don’t know how much you use AI. You use it throughout the day, but it’s fragmented. A prompt here, a quick question there, a code snippet, a research session. Each instance feels necessary. But the aggregate effect—the total time, the total dependency, the total number of decisions you’ve outsourced—is invisible to you.

This is why most people have no idea whether their AI use is getting better or worse. They can’t see the patterns because they’re not measuring anything.

A simple log changes this. Not to judge yourself or restrict yourself. To see clearly. And once you see, you can actually make decisions about whether your use is serving you.

What a Basic AI Log Looks Like

You don’t need something elaborate. A spreadsheet or note document is enough. For each session where you use AI, log:

  • Date and time (or just time of day)
  • What you used AI for (be specific: “drafted social post,” not “content”)
  • How long (ballpark is fine: 5 minutes, 30 minutes)
  • Output (did you ship this, or was it just thinking?)
  • Could you have done this without AI? (yes/no)

That’s it. Five columns. Takes 30 seconds per session to log.

Do this for two weeks. At the end, look at the data.

What the Log Reveals

After two weeks, patterns emerge.

Your actual usage volume: You probably use AI more than you think. When you see “15 AI sessions today,” you realize the fragmentation cost. When you see “the entire afternoon was AI prompting,” you understand the context-switching loss. The data makes invisible patterns visible.

Your use-case distribution: What are you actually using AI for? Research? Writing? Code? Decision-making? Brainstorming? See which activities are AI-heavy. For each category, ask: should this be AI-heavy, or am I defaulting to AI when I should be thinking?

Your dependency patterns: Are there tasks you’ve completely outsourced? Can you still do them without AI? This is where atrophy becomes visible. If you log “wrote three emails with AI, did zero without AI,” you’ve found a place where you’re losing capability.

Your decision patterns: How many decisions did you delegate to AI this week? “Ask AI which framework to use,” “Let AI decide whether this design works,” “AI recommended X.” What are the consequences of these delegated decisions? Are they good?

Your speed tradeoff: You probably use AI for things you could do yourself slower. Is that tradeoff working? Are you getting more done because you’re faster on routine tasks? Or are you just doing more routine tasks and neglecting important thinking work?

📊 Data Point: People who kept a two-week AI use log showed significantly increased awareness of patterns they couldn’t see in real-time. 70% reported changing their AI use patterns after seeing the data.

💡 Key Insight: You can’t change what you don’t measure. Awareness is the first step.

Using the Log to Make Decisions

After two weeks of logging, you have data. Now you can make conscious decisions.

Question 1: Is this the right volume? Look at total sessions per day. Are you using AI more than you expected? Is the volume helping or hurting your deep work? If the volume is too high, design your next two weeks with a usage target. Aim for lower volume. Log again. See if you can maintain lower usage while still getting work done.

Question 2: Are these the right use cases? Look at your use-case distribution. For each category, ask: should I be using AI for this? Is this helping me work better, or am I just using AI because it’s available?

Some categories you’ll want to cut back on. “Writing every email with AI” might become “write 80% myself, AI-draft when stuck.” “Every decision with AI” might become “use AI for synthesis, but I decide.”

Question 3: Where am I losing capability? Look for activities where you’ve gone 100% AI. If you see “zero code written without AI” or “all writing is AI-generated,” that’s where atrophy is happening. Design a rebalancing. “This week, I’ll write 20% of my code without AI” or “I’ll write one article from scratch.” Rebuild capability in areas where you’ve gone fully dependent.

Question 4: What’s the quality tradeoff? Not just speed. Quality. Is the work shipping with AI help noticeably different from work you do yourself? Better? Worse? Different? Once you see the pattern, you can make conscious choices about where the difference matters.

Making It a Habit: Sustainable Logging

Logging doesn’t need to be perfect. It needs to be consistent.

Make it fast: 30 seconds per session. If it’s taking longer, you’re being too detailed. Simplify.

Do it immediately: Log right after using AI. Don’t accumulate sessions and try to log later. You’ll forget. Immediate logging captures the actual experience.

Pick a tool that lives in your workflow: Phone notes, spreadsheet, a quick bullet journal. Whatever you’ll actually use. It’s not about the format. It’s about consistency.

Review weekly, not daily. Don’t obsess over individual sessions. Once a week, look at the aggregate. See the patterns. Adjust if needed.

Log for four weeks, then decide. After the first two weeks of baseline, adjust your usage patterns based on what you learned. Log for two more weeks of the new pattern. See what changed. This tells you what’s actually within your control.

What This Means For You

Start logging tomorrow. Just five columns. Five questions per session. Do it for two weeks.

You’ll be surprised by what you see. Most people discover they use AI more than they thought, that they’ve outsourced thinking they didn’t realize they’d outsourced, and that they have more control than they think.

Once you see the data, you can make real choices. Not motivated by shame or restriction. Motivated by clarity. You’re not trying to use AI less. You’re trying to use it consciously.

And that’s where control lives.

Key Takeaways

  • A simple AI use log (date, task, duration, output) reveals patterns invisible in real-time.
  • After two weeks, you see: actual usage volume, use-case distribution, dependency patterns, decision patterns, and speed tradeoffs.
  • Use the data to answer: right volume? right use cases? capability loss? quality tradeoff?
  • Logging itself is fast (30 seconds per session), immediate (right after use), and sustainable (weekly review).
  • Awareness is the prerequisite for change. You can’t control what you don’t see.

Frequently Asked Questions

Q: Is logging itself another form of friction/obsession with AI use? A: Not if you keep it simple and weekly-reviewed. If you’re obsessing about individual sessions daily, you’ve turned it into anxiety. That’s not the goal. The goal is quarterly or monthly patterns, not daily optimization.

Q: What if I discover I’m using AI way more than I thought? A: That’s valuable information. Now you can decide: is that level appropriate, or do you want to reduce? But at least you know. Most people never discover this because they don’t track it.

Q: How long do I need to keep logging? A: Ongoing works best, but it doesn’t need to be forever. Four weeks of baseline plus adjusted usage gives you solid data. Then you can log periodically—a week every quarter—to catch changes.

Not medical advice. Community-driven initiative. Related: AI Session Planning | The Intentional AI Use Protocol | Mindful AI Use