Judge human judgment by how it handles uncertainty

TL;DR: Cleo Eleftheriades compresses ten kinds of judgment into an eleven-second card sequence. I would use the list to design a decision exercise, not to predict which abilities AI can never acquire. The useful standard is how a person handles uncertainty and takes responsibility when the answer changes.

This share is for leaders trying to describe the human contribution more precisely than critical thinking. The cards include questioning premises, evaluating evidence, considering consequences, mobilising people and adapting under uncertainty. The ranking is the creator's opinion. Original reel and caption.

The audio is a countdown; the detail is written on the cards. I checked the caption and frames covering all ten entries. Around 0:03, the sequence reaches metacognition; around 0:06, it moves through evidence and context; around 0:08, it arrives at adaptive judgment and responsibility.

Put a decision under changing evidence

Here is the hypothetical exercise I would use with a technical lead. A team must decide whether to roll out an agent-assisted process to another department. Give the lead a small evidence pack with promising results, uneven coverage and one unresolved failure.

Ask for a recommendation, but also ask which assumption carries the most weight and what new information would change the decision. That makes the reasoning inspectable before the outcome is known. A confident recommendation without a revision condition is harder to distinguish from commitment to a position.

Then introduce a new fact that challenges the chosen assumption. Perhaps the successful cases came mostly from one unusually clean data source. The interesting response is how the person revises the rollout boundary and communicates the change, not whether they defend their initial answer elegantly.

This is my proposed exercise, not a validated test of leadership or a human-versus-model benchmark. It makes several items in the reel observable without assuming they belong permanently to one kind of intelligence.

Accountability begins after the answer

A decision also needs somebody to organise what happens next. If the recommendation is a narrower pilot, who owns the acceptance condition, the unresolved evidence and the communication to the affected team? A good analysis can still fail as leadership if nobody acts on it.

I would assess the follow-through alongside the reasoning. Did the decision owner make the uncertainty visible, invite the relevant expertise and return when the agreed evidence arrived? Those behaviours matter even when an AI system helps prepare every document.

The reel's vocabulary is useful, but its ordering should not become a universal skills hierarchy. Taste, technical knowledge and coordination can matter differently across tasks. Nor does labelling a capability human prove that a person demonstrated it in a particular decision.

Try the exercise on a low-risk project choice. Keep the initial recommendation and the revision side by side. The change between them can reveal more about judgment than a polished statement of principles written after everything worked out.

Judge the decision-maker by how they update the call and carry responsibility beyond it.