Testing is a deeply human experience.

Those quiet moments when some part of a solution reveals itself. The maddening frustration of peering into the abyss of the unknown looking for problems.

Shared moments with the team taking that journey together.

Those are the parts of being human that can’t be encoded in a process or training data.

Those are the parts that make testing beautiful.

When testing is done well (and sometimes not so well), it entwines experiments, learning, and emotions to help us create meaning.

Meaning in a messy, socially inconvenient investigatory exercise that thrives on uncertainty.

So when I see the AI Confidence Bros have discovered testing, my initial reaction is “god help us”, but then eventually turns to laughter.

They don’t get it.

It’s all versions of the same tired fantasy that if we just capture enough human behavior, codify enough process, and feed enough examples into the machine, we can create something better than the humans who built the mess in the first place.

No, we can’t.

Human behavior is not deterministic. It is inconsistent, political, emotional, contextual, performative, and frequently ridiculous.

People who live with the consequences of having to do testing that matters know that.

Real testers have been using “about,” “estimated,” “approximately,” “based on what we observed,” and “here is what we don’t know” for decades.

Because courage in the face of uncertainty is what builds trust.

So when this group fails, as they all inevitably do, to create their human shaped robot that they can put a human into to act like a robot shaped human, we can add a line to the notes left on the table for them.

“Congratulations. You’ve discovered testing.”

End note: I have to say a massive thank you to Emily Bender and Alex Hanna for not only inspiring this post, but for their tireless and thankless work in pushing back against the tidal wave of AI psychosis that has a grip on society.