Holding a hammer, everything looks like a nail
With AI in hand, you want to give it every task. It's like holding a hammer: everything looks like a nail. But not every task is worth it.
TWO QUESTIONS
Two filter questions
Before you start, ask two questions. One: if AI gets it wrong once, what happens? Two: is it easy to check its work?
DO IT NOW
AI's sweet spot
Drafting an email, summing up a meeting, naming a new cake. If it's wrong, you fix a few words, and one glance tells you if it's right. This is AI's sweet spot.
CAREFUL
The red zone: AI helps, people decide
But calculating salaries, diagnosing illness, signing contracts for you: one mistake costs money or health, and checking it almost means redoing it from scratch. Here, AI should only help, and a person must decide.
NO AI NEEDED
When a formula is enough
Some tasks don't need AI at all. Calculating tax with a fixed formula? A spreadsheet is enough: faster, cheaper and right every time.
RECAP
Put it all in one table
Put it all in one table: small mistakes that are easy to check, do it now. Big mistakes that are hard to check, avoid. In between, let AI draft and a person approve.
PART 3
AI still makes mistakes sometimes
You've picked the right tasks. But AI still gets things wrong sometimes. How do you design so those mistakes don't cause harm? See you in part 3.
This article is based on the video Which tasks are worth using AI for from the Mark học AI channel. Watch the video (in Vietnamese) to see the animations.