ARNIANO

Insights · Andrew Clayton

Why Lateral Thinking Matters More in the Age of AI

I first encountered Edward de Bono as a young designer in England. One apparently nonsensical question about a clutch taught me why deliberately breaking the frame can matter more than producing another competent answer inside it.

I first came across Edward de Bono in England in about 1982. He was already well known, and we studied his ideas at Technical College. At the time I was a young designer, and lateral thinking was not something I encountered as an abstract management concept. We actually used the methods. The appeal was practical: if a group had become trapped inside the obvious way of looking at a problem, deliberately changing the frame could force everybody to see it differently.

One particular discussion has stayed with me for more than forty years. A group of us were talking about car clutches. I asked whether changing the polarity of a fluid could change the force required to rotate a propeller or rotor within it. I was exploring a possibility rather than presenting a developed invention; the physical mechanism would still have needed to be understood and tested. What I remember is the reaction. The conversation stopped. The people in the room with the strongest scientific grounding looked at one another and, for a moment, everyone had to reconsider the framework within which we had been discussing the problem.

That was the important part. The question did not need to be correct to be useful. It broke the existing pattern. Once the familiar assumptions had been interrupted, people were forced to ask what actually determined the resistance, which variables were fixed only because we had assumed they were fixed, and whether the mechanism itself needed to be approached in the conventional way. The triggering effect was more valuable than the literal suggestion.

I have carried that example with me because it captures what I think de Bono understood particularly well. Expertise is essential, but expertise creates patterns. The more we learn about how something normally works, the more efficient we become at moving through the established pathways. That is usually an advantage. The difficulty arises when the pathway itself is the constraint. At that point, more expertise applied in the same direction can simply produce a more sophisticated version of the existing answer.

Lateral thinking attempted to make the interruption deliberate. Reversal, provocation, random entry and the deliberate suspension of judgement can appear artificial, but that is partly the point. They create permission to introduce an idea before it has been filtered through the normal requirement to be sensible. Most such ideas will lead nowhere. Occasionally one changes the question, and changing the question can be more valuable than improving the answer.

This has become more interesting again with generative AI. AI is reducing the time and cost involved in assembling a conventional analysis, although the result still requires examination. An AI system can summarise the established literature, build a framework, identify competitors, list risks, draft a plan and produce the conventional alternatives in a fraction of the time that these tasks once required. That is an extraordinary gain in productivity, and I use it extensively. But it also changes where human value sits.

If everybody has access to broadly the same bodies of knowledge and increasingly capable models, conventional competence will become less differentiating. The danger is not that the answers will be poor. The danger is that they will be good and increasingly similar. A market can fill with well-structured strategies, polished presentations and reasonable recommendations that all arise from essentially the same assumptions.

In that environment, originality becomes a fresh discipline. The valuable question may no longer be ‘what is the accepted framework for this problem?’ because the framework can be generated instantly. The more interesting questions become: which assumption should we challenge, what if the apparent constraint is not fixed, what would somebody from another industry see here, and what happens if we deliberately reverse the relationship? Those are recognisably lateral-thinking questions.

AI can actually make this easier if it is used in the right sequence. Let the machine establish the conventional position quickly. Ask it what the standard models say, what competitors normally do, what the likely risks are and which assumptions are embedded in the existing approach. Then use that foundation as something to push against. The purpose is not to reject knowledge but to avoid becoming trapped by it.

There is also something reassuring in this for people who worry that AI will make human thinking less important. I suspect the opposite may happen. The more cheaply we can obtain a competent answer, the more valuable it becomes to recognise that the problem has been framed incorrectly. Curiosity, contradiction, intuition and apparently irrelevant experience become useful precisely because they can interrupt the pattern that the machine reproduces so efficiently.

I still think back to that clutch discussion. Whatever the merits of the suggestion about the fluid, it made a room of technically trained people stop applying what they already knew and look again. In an age when established knowledge can be assembled almost instantly, the ability to create that interruption may be more important than ever.

All insights