Hey! Ankur here, and this is the 14th edition of Lazy AI — 5 mins of reading to help you stay ahead of the AI curve.

There’s something about how we use AI that nobody says out loud. But after you read what I have to say, you’re going to nod and then feel slightly comfortable.

Think about how you actually use AI end-to-end. Not just for the first draft, but for the full thing.

If you’re an engineer, maybe you use it to write documentation. If you’re a product manager, maybe you use it to throw together a rough prototype. If you’re a marketer, maybe you ask it to crunch some data. If you’re a financial analyst, maybe you use it to draft an email that doesn’t sound passive-aggressive.

Notice anything?

Every single one of those use cases has the same pattern. When you’re using AI for end-to-end tasks, you’re using it for something outside your core skill. The thing you’re not trained for. The task that’s adjacent to your job, not the job itself.

That’s not a coincidence. It’s deeper.

If this article is adding value, do consider subscribing for periodic emails. I write such that you understand AI at its core :)

The people who are genuinely excellent at what they do — the engineers who’ve been building for 15 years, the analysts who can spot a bad balance sheet in seconds, the writers who’ve spent a decade developing a voice — they don’t hand their actual work to AI.

Yes of course, they use it for a first draft. A quick outline. A sanity check. But end to end? Not a chance.

Why?

Because there’s a difference between pattern and taste. And right now, AI only has one of them.

You see, a Pattern is what AI does extraordinarily well. It has consumed more human-generated text than any person could read in a thousand lifetimes. And when you ask it something, it predicts the most plausible, well-formed response based on everything it absorbed. The AI is not really thinking. It’s predicting. Extremely well.

AI works on probability, where all it does is predict the next word based on how it’s trained. If you’re keen to know more, this was the first article I’d written in the LazyAI newsletter.

Now, here’s what a usual day-to-day task (given to AI) looks like in practice:

Ask AI to write a solid brief for a marketing campaign. It’ll give you something competent — the right sections, the right tone, the right structure. Because it has seen ten thousand briefs and knows what they look like.

Now ask your best marketer to review it.

She’ll change things. Not because the brief is wrong. But because she knows — from years of watching campaigns live and die — that this particular client panics when you lead with awareness metrics, that this product needs to be sold on feel not feature, that the brief needs a line in it that nobody in the room expects. She can’t fully explain it. She just..knows.

That’s taste.

Taste is what you build after you’ve been wrong enough times in exactly the right ways. It’s the judgment that sits behind your instincts — the thing that tells you something is off before you can articulate why. AI doesn’t have it. Not because it’s not smart enough. But because taste isn’t learned from reading about the world. It’s learned from physically going through it.

Loosely put, AI knows what’s right. It doesn’t know what’s wrong. Yet. And that disruption (called Artifical General Intelligence or AGI), will take time.

Let that sink in..

Every AI headline measures the wrong thing.

“AI passes the law exam.” “AI beats doctors at diagnosis.” “AI writes better than junior engineers.”

All measuring AI against the average. Against the pattern. And of course AI is good at pattern. I mean, that’s literally what it is!

The real benchmark is not whether AI can do the task. It’s whether the person best qualified for that task would trust AI to do it end-to-end. Without a rewrite. Without a second look.

And right now? The answer is almost always no.

See, I’m not here to tell you what AI to use or how to use it — that’s your call. But ask yourself: What do you use AI for, and what do you never fully hand over?

That gap is where you come in. And the day the best people in every field stop needing to correct AI’s output — that’s when we’ll have something worth calling true intelligence.

Till then, if you’re above average at your job, AI will not take it.

If this newsletter was an Instagram reel, I’d say “Send this to your boss to tell them why they shouldn’t replace you with AI” 😉

But this is more for you than for them..

That’s all for today! If this got you thinking, share it with someone who uses AI every day and tell them that they’re safe..for now.

It takes a lot of time to research and write this. Sharing it would mean a lot to me 🙂

See you next time..

Cheers,

Ankur