An afternoon with philosopher Dr Peter Ellerton left little room for small talk.

6-minute read.

Dr Ellerton, founding director of the University of Queensland's Critical Thinking Project, was the first speaker in Sky Discovery's What Matters series — an invitation-only lunch bringing senior lawyers together for conversations that don't usually happen over a desk. His topic: what he calls "principled rebellion," the case for refusing to surrender human reasoning to the convenience of AI. Hard to think of a more relevant subject for a room full of lawyers in 2026.

What follows are seven ideas from that conversation worth carrying back to the office — particularly for a profession where judgment is the product.

1. Being smart isn't the same as thinking well

Peter opened with a distinction that's easy to nod along to and harder to sit with: intelligence and critical thinking aren't the same thing. You can be highly intelligent and reason poorly, and you can think rigorously without being the smartest person in the room. Critical thinking, as he defines it, is the study of the quality of our inferences — the decisions, recommendations, and conclusions we reach — and the factors that affect that quality.

The useful part for professionals: unlike raw intelligence, this is learnable. Critical thinking isn't a fixed trait you either have or don't — it's a set of skills and habits, which means it's also a pathway to expertise rather than a by-product of it.

2. Your reasoning has to be visible

Peter's simplest and most exportable tool came wrapped in one sentence. After any recommendation — What should we do here? What's your read on this? — add one more question: "And what's your argument?"

It sounds obvious. In practice, it's rarely asked, and even more rarely answered well. Reasoning that can't survive that question probably shouldn't survive contact with a client, a partner, or a bench either. Making thinking visible isn't just a courtesy to whoever's listening — it's what allows anyone, including the person doing the reasoning, to test it, refine it, or catch where it's gone wrong.

3. Structured groups outperform smart individuals

Peter calls it the assembly bonus: when a group shares its reasoning openly and properly tests each other's thinking, it reliably outperforms even its most capable individual member. Not because groups are inherently wiser, but because the structure — visible arguments, genuine challenge, a shared standard for what counts as a good reason — does the work that raw talent alone can't.

The implication for legal teams is a practical one: seniority and intelligence matter, but they're no substitute for a culture where reasoning gets said out loud and tested, rather than assumed and deferred to.

4. Precision in language isn't pedantry

Peter set up a simple scenario. A goanna is clinging to a tree stump, always keeping the trunk between itself and a dingo circling below. The dingo walks a full loop around the stump. Question: has the dingo gone "around" the goanna?

The room split. Some said yes — the dingo traced a complete circle in space, so of course it went around. Others said no — the dingo never saw the goanna's other side, so from the goanna's point of view, nothing "went around" it at all. Both answers are reasonable. Nobody disagreed about what actually happened; they disagreed about what the word "around" means.

That's the point Peter was making. Small wording choices — saying a deadline falls "around" the end of the month rather than naming the date, or treating "accurate" and "precise" as interchangeable when they're not — aren't a style preference. Left undefined, they let two people agree on every fact and still walk away with opposite conclusions.

For a profession built on the exact wording of contracts, evidence, and instructions, that risk is a familiar one. An undefined key term isn't a minor imperfection. It's a disagreement waiting to happen.

5. Facts are necessary, but they don't finish the job

Even with every fact on the table, Peter argued, big decisions often remain underdetermined. His example: when two Olympic athletes tie for gold, should there be no silver medal, or should the next athlete still receive one? The facts of the tie don't answer the question — what decides it is the quality of the argument and the values used to weigh the options.

The same is true of most judgment calls professionals are paid to make. Complete information is rarely the constraint. The quality of the reasoning applied to that information is.

6. AI doesn't lie — it's indifferent to truth, which is a different problem

This was the sharpest turn of the afternoon. Peter borrowed a distinction from philosopher Harry Frankfurt: a liar at least cares about the truth, because they're positioning themselves against it. A "bullshitter," in Frankfurt's technical sense, doesn't care whether a statement is true or false — only whether it's convincing. By that definition, Peter argued, AI output is truth-indifferent by design: sometimes right, sometimes wrong, but not answerable to the difference either way.

That's not an argument against using AI — Peter was quick to note the flip side. "AI can't teach you how to think critically," he said. "It can only improve you if you already can." The tools amplify whatever quality of thinking you bring to them. They don't supply it. Which means "the AI got it wrong" is a weak answer for a professional to give a client — the accountability was never the model's to hold in the first place.

7. None of this is optional anymore

Put together, the case is straightforward: if AI can produce fluent, confident output that's sometimes false, the professionals who can show their reasoning — who can answer "and what's your argument?" without flinching — are the ones clients will keep trusting. That was true before AI. It's simply less optional now.

What Matters is a Sky Discovery conversation series for senior lawyers. Dr Peter Ellerton's talk in Brisbane was the first of three events across 2026 — Sydney and Melbourne to follow.