We handed our own AI a photo of this ordinary wall clock and asked one simple question: what time is it?

The answer came back instant and confident — and it was wrong. Not a little wrong. The model never really read the hands; it reached for the one time nearly every clock in its training is set to, and guessed that instead.

It took our AI five rounds of analysis and roughly 30,000 tokens to finally read it — and its first guess was still wrong. A child does it in about a second.

TLDR: AI isn't smart or dumb in a straight line. It's spectacular at some surprisingly hard tasks and clueless at some surprisingly easy ones — a pattern researchers call the "jagged frontier." Once you can see its shape, you stop being surprised and start being strategic about what you hand the machine.

Here's what makes it land: that same model passes the bar exam, drafts diagnoses doctors agree with, and writes production code. On the ClockBench benchmark, humans read analog clocks with 89% accuracy — the best AI managed 13%. It aces the hard thing and flunks the easy thing in the same breath.

That's not a flaw. It's a map. Every place AI stumbles is a place you now know to keep a human in the loop — and every place it soars is a place you can lean in hard.

The next time AI dazzles you, remember the clock. The most useful thing you can know about a genius is exactly where it's blind — because that's the spot where you are still the expert.

This guide from Deel breaks down how to build one global hiring system. You’ll learn about assessment frameworks that scale, how to do headcount planning across regions, and even intake processes that work everywhere. As HR pros know, hiring in one country is hard enough. So let this free global hiring guide give you the tools you need to avoid global hiring headaches.

Download the free guide today