My past mentee (2 YoE Engineer) is worried.

There are many different wordings to this quote, to cite one.

“Software Engineering Will Be Automatable in 12 Months” - Dario Amodei, Anthropic CEO.

Setting aside obvious vested interest in such a quote being believed, will it?

Current ARC-AGI 3 result says otherwise.

Since the new AGI benchmark’s release in 2026-03-26, it’s almost a month now, no one has managed to make their AI agent score over 1%.

1% out of 100, not even there yet.

What’s going on?

The new AGI 3 benchmark is testing fluid intelligence. Navigating an environment without being given a rule, interacting with environment, figuring out the rules, then winning.

1 word: Intentionality.

My mentee said “I don’t think it needs to get to the level of AGI (100%) to replace us as a software engineer.”

To be fair, he has a point. My reply, which probably resonates with about 1/2 of professionals on linkedin:

“For my own project, I mostly don’t write or edit a single line of code. That part is true.

I still need to decide what to build, decide architectural direction, catch agent taking shortcuts, or getting hyper-fixated on a part and neglect the whole.

I don’t believe in ‘release an agent swarm to build an app overnight’. At least for the kind of things I build, I need to babysit for Claude code to code correctly.”

Whether my experience generalises is beside the main point.

The main point I told him was:

“Not an easy time to know what to expect. If you think you have no idea of the future, I and everywhere you apply to is about as clueless as you are.”

I believe no one really knows what’s going to happen. Everyone who claims otherwise, like Dario Amodei, is just pretending.

In light of this, the greatest resource isn’t just the fluid intelligence to direct LLM, it’s navigating uncertainty with intentionality. That is precisely the fluid intelligence skill that ARC-AGI 3 tries to measure, and at least at this point in time, still very human by any measure.

How to navigate the world. How the world works. What you want. Where opportunities are. No one optimises that learning for you. No one tells you what you’re supposed to be doing.

But you don’t need anyone to tell you. As a human, you’ll outdo 2026 AI 100 to 1.

Certainty was never going to happen. Intentionality is what sets you apart, in more ways than one.