When I say the 'stuff in the middle', I'm talking about what happens between having a question and having an answer to that question. We often talk about the importance of crafting the correct question - or understanding the problem. And then we like to have an answer to our question, but sometimes realise that actually there isn't just one answer, but rather a range of possibilities - how do I know which to choose? So, the stuff that happens between the Q and the A, can be - neigh is - crucial to the quality of both.
When we outsource the 'stuff in the middle' as we often do in various ways to make finding an answer easier and quicker, what parts of the process that used to happen on the way to that answer do we accidentally remove? And what if some of those parts were actually where much of the value was created?
Last week, we looked at how AI is changing the way we search for information. But finding information is only part of the story. What about the 'research', exploration, analysis, and thinking that turns information into something useful? Here's what I'm thinking .....
I remember being told at school so many times to “show your working” when solving a maths or science problem. Getting the right answer was important, but demonstrating how you arrived at it seemed to be even more important.
And there was a good reason for that. The process helped us develop our understanding and the skills to tackle the next problem, which perhaps would be a little different, more complex, and more difficult.
And outside the classroom, many problems don’t have a single right answer anyway. We have to work towards a solution through investigation, reasoning, and judgement. There are many variations on the old adage that 'it's the journey that's important, not just the destination', and from a Q to A perspective, the journey matters because it changes not only what we know, but how we think about the problem.
If we think about how we might traditionally approach a business problem. We start with a question, gather information, explore different perspectives, identify contradictions and inconsistencies, challenge assumptions, and perhaps discover something entirely unexpected. And while doing that, at least 3 important things can happen:
And through all of this, we develop our own understanding and capabilities. We become better equipped to tackle the next challenge. That’s quite a lot of value being created in what might otherwise look like an inefficient process.
AI can be extraordinarily useful in accelerating this journey. It can find and summarise information, compare arguments, suggest alternative explanations, identify connections, and help us explore unfamiliar territory. Why wouldn't you just want to use that all the time?
In some respects, we’ve been here before. When Wikipedia became popular, people questioned its reliability, its biases, and whether we should trust information written by people whose expertise we couldn’t always verify. Yet Wikipedia has become an invaluable starting point for exploring unfamiliar subjects. We learn to recognise its limitations, follow its references, and exercise our own judgement about what to trust. It's not perfect, there are biases, it's a bit like everyday life
AI presents a similar challenge, but with an important difference. It doesn’t just provide information; it aims to interpret, combine, and present it as a coherent, and often very convincing answer. And that offers a rather tempting shortcut from Question to Prompt to Convincing Answer, and then you're Done.
While there is a danger that the answer is wrong - and it certainly does happen - the bigger danger and problem is that we might bypass the thinking, questioning and exploration that would have helped us understand whether the answer was appropriate in the first place. And so, there is an important difference between using AI to do our thinking for us and using AI to help us think better.
This is becoming an increasingly common question in professional services, including consulting. Clients quite reasonably want to understand whether AI has been used in work undertaken on their behalf. But perhaps the more useful question is 'how was AI used?' There's a lot of difference between accepting an AI-generated answer and using AI as part of a disciplined research process - investigating sources, testing assumptions, challenging conclusions, exploring alternatives, and applying professional judgement.
Both involve AI. But the quality of the process, and potentially the outcome, can be very different. Of course, questions about confidentiality, data security, and accountability remain important regardless of how AI is used.
Sometimes we really do just need an answer; there’s little value in spending half an hour researching a straightforward conversion that a calculator can perform in seconds. But even then, there’s a distinction between using a tool to save time and becoming so dependent on it that we lose the underlying skill (how much change do I get from a $20 note when I buy something for $13.37).
And when we’re dealing with more complex questions - particularly those involving people, organisations, uncertainty, and change - understanding how we arrived at an answer may be every bit as important as the answer itself.
But there’s another danger of outsourcing the 'stuff in the middle': we may not recognise that we need to understand more than we do. A convincing answer can give us a comfortable sense of certainty, even when the underlying evidence is incomplete, the assumptions questionable, or important alternative perspectives have not been considered. Critical thinking requires us to remain willing to question what we think we know, to recognise the limits of our understanding, and to live with some uncertainty. That isn’t always comfortable, and like any capability, if we stop exercising it, we risk losing it.
Is AI the villain here? It's more likely that its nature combined with our human nature is a potentially dangerous mix. But used thoughtfully, AI can make our exploration considerably richer, faster, and more productive. But it can also make it remarkably easy to settle for an answer before we’ve really understood the question. We need to keep ourselves mentally fit enough to tolerate uncertainty, to challenge comfortable answers, and to recognise when we still have more to learn.
So perhaps, before asking AI for an answer, we should ask ourselves: 'do I just need the answer, or do I need to understand?'
And perhaps more importantly: 'how will I know if I haven’t understood enough?'
Because sometimes, the stuff in the middle is where the real value lies.
For more background, see SEO Isn’t Dead, But It Is Different… and Curiosity, Learning, Knowledge and Agency.
Ad Futurum
Graham
/kcl-goes-green/HeaderLogo-60x300-ID-e065fb7a-094d-4fc6-d8c1-f4f88e7523f5.png)