Aristotle vs. Artificial Intelligence

Every week brings another announcement about AI agents that can plan, decide, and execute on our behalf.

This is no longer just about better answers. It is about delegating action.

Which makes a 2400-year-old question newly urgent:

What is judgment actually made of?

For Aristotle, the highest form of practical intelligence was not technical skill.

It was phronesis: practical wisdom.

The ability to see what this situation, with these people, at this moment, actually calls for.

Phronesis is not just rule-following.

It is the cultivated ability to recognise when the general rule does not fit the particular case.

You can know every ethical framework and still lack it.

You can be brilliant at logic and still not know when to speak, when to wait, when to challenge the rule, or when the “efficient” answer is exactly the wrong one.

Phronesis is built slowly: through experience, mistakes, habit, character, and paying attention to reality when it refuses to fit neatly into a model.

Now compare that to contemporary AI.

AI systems are extraordinary at generalisation. They extract patterns from huge amounts of data, recognise similarities, and generate fluent recommendations.

And increasingly, they are being asked not just to answer — but to act.

• Schedule the meeting.
• Rank the candidate.
• Approve the workflow.
• Flag the customer.
• Write the message.
• Escalate the case.

That is where the philosophical question becomes practical.

Because real judgment often begins with noticing:

• This case is different.
• This person is not the average user.
• This decision has consequences the model cannot feel.
• This rule was made for normal cases, and this is not one.

That is the gap.

AI can support judgment.

But it does not possess phronesis.

• It does not have character.
• It does not learn through moral responsibility.
• It does not carry the weight of being answerable for what happens next.

So the question for builders, leaders and policymakers is not only:

“Can AI reason?”

It is:

“When the right answer is not the average answer, who is still in the room?”

Aristotle would probably remind us that practical wisdom cannot simply be downloaded, fine-tuned or prompt-engineered.

It has to be formed.

And that should worry us.

Because if we automate everything except judgment, we may end up with a world full of fast, fluent, optimised answers — while slowly weakening the human capacity to recognise when optimisation is not wisdom.

In an AI-saturated world, the edge will not go to whoever automates the most.

It will go to whoever still knows when the average answer is the wrong one.