The Attribution Gap

When AI decisions fail, who's accountable?

by Sam Rogers
3 min read
video
governance
founding-partner
accountability
risk-management

It's Tuesday, 9:23 AM, and something goes wrong.

Watch the Video

Watch on YouTube →

The Timeline of a Failed Decision

A decision gets made. It was informed by AI. The output looks solid—clearly reasoned, confident. The decision-maker confidently acts upon it and moves on to the next thing.

But by noon, everyone is well aware that decision turned out to be wrong. Painfully wrong.

Now it's 1pm, and somebody has to answer for it.

1pm: Look at the Employee

So you look at the employee. Hey, they followed the process. They used the tool the way they were trained to use it. They didn't do anything negligent—they did what most anyone in their position would do. They trusted a confident output.

2pm: Look at the AI

By 2pm you look at the AI. It produced exactly what it was asked to produce. It wasn't broken. Malfunctioning? ...no. It just didn't flag that its answer was built on an assumption that nobody had questioned.

3pm: Look at the Organization

By 3pm, you look at the organization. The policy was in place. The training? Completed. The access controls are all configured. Everything that was supposed to be done was—yep—done. Good!

And yet, the outcome is still very, very not good.

4pm: The Question That Matters

4pm, it's real. No more wiggle room. Now... who's accountable?

This Is the Attribution Gap

I'm Sam Rogers, founder of PAICE.work.

When AI is used in a decision like this, and that decision fails, the accountability chain disintegrates on impact. Not because people are being evasive, but because nobody was measuring the one thing that actually matters: what did the human do with the AI output?

  • Did they verify it?
  • Did they question it?
  • Did they maintain judgment, or did they defer?

Right now, most organizations have no way to answer these questions.

The Attestation Problem

You can prove someone completed training. You can prove they used the tool. You cannot prove they exercised due diligence in how they collaborated with that tool.

And that's the gap that matters when the auditor asks. When the regulators ask. When it's 4pm and leadership is asking you.

It's not "did they use AI?" It's "did they use AI right?"

And if you can't demonstrate the difference, that's not a training problem. That's an attestation problem.

How PAICE Closes the Gap

PAICE.work measures the behavior that closes this gap. It's not what people know about AI—it's what they do with it.

If you're responsible for AI governance and you don't have an answer to the attribution question? Please, let's talk.

Because right now your organization is vulnerable. You're vulnerable. And you simply don't need to be by the time next Tuesday rolls around.

Get Involved:

Curious but short on time?

Take the 3-minute PAICE Pulse — a quick confidence check that maps how you see your own AI collaboration posture. No login required.