LANDMARK · 5 MIN · ORIENT
Automation and autonomy
After this landmark, you can place any AI system on the line from fixed automation to full autonomy, and locate where a human approval sits.
Automation and autonomy aren’t the same thing. A fixed automation does exactly one scripted thing every time — a mail rule that files receipts, a script that renames files. An autonomous system decides its own next steps: it looks at the situation, chooses an action, sees the result, and chooses again. Most real AI sits somewhere on the line between the two. More autonomy means more usefulness, because the system handles cases you didn’t script — and less predictability, because you can no longer read off exactly what it will do. The question that matters most is where a human approval sits before an action takes effect.
This is the canonical concept. It stays the same across learner lenses so personalization never changes the underlying facts.
◌ EVERYDAY LENS
What this looks like for you
A reminder that fires on a schedule is fixed automation — safe and boring. An assistant that “handles” something for you is choosing steps on its own; check what it can do before it does it, not after.
MAKE A DECISION
You’re choosing how an AI tool should send follow-up emails to customers. What’s the safest first design?
CARRY THISTake one task you’d like AI to handle and mark the single step where a human should approve before it acts.