The Thinking Is in the Process

A few days ago, I had a discovery conversation with a prospective client. We talked about their business, the challenges they were facing, the people they wanted to develop and some of the possibilities for working together.

I listened, asked questions and took notes (hand written not typed) and I didn't record the conversation, although I could have. I could also have uploaded a recording to AI afterwards and asked for a summary, a list of key themes and perhaps even a draft proposal. Which would probably have been faster.

Instead, I followed a rather more circuitous route, and it got me thinking about something we frequently discuss in relation to AI: discernment.

The first round of thinking

When taking notes during a conversation, you are constantly making choices. You can't write down everything someone says (and wouldn't want to). You are likely listening to their words, paying attention to the meaning behind them, making connections and deciding what seems important enough to capture.

Sometimes I'll write down an exact phrase because the language feels significant. Sometimes I'll summarise several minutes of conversation in a few words. Occasionally, I'll make a note of something that doesn't quite fit, because I suspect it might matter later.

All of this happens while I'm continuing to participate in the conversation. I'm listening, interpreting, questioning and selecting. That's discernment in action, although I don't generally think of it that way at the time.

Then I left it alone

After the meeting, I put the notes aside for a couple of days. I didn't consciously spend that time analysing the conversation. I simply got on with other things and when I returned to the notes, I noticed that some things stood out more clearly than they had during the meeting. A few connections became more apparent, and I found myself reconsidering the relative importance of different points.

Perhaps the distance helped. Perhaps I was simply approaching the material with a different frame of mind. Either way, I was engaging with it again.

And that matters, because each time we revisit an experience, we have an opportunity to notice something new, make fresh connections and deepen our understanding.

Thinking out loud

The next step was to read my notes aloud to AI.

I could have photographed the pages and uploaded them, but I chose to speak instead. And as I read oput loud I realised I didn't simply read what I'd written. I expanded on certain points, changed the sequence of others and explained connections that weren't explicit in my notes. Occasionally, I remembered something from the original conversation that helped put an observation into context.

In other words, I was interpreting my own interpretation.

The first time, I'd taken a conversation and turned it into notes. Now I was taking those notes and turning them back into a narrative. That required another round of judgement.

What mattered? What belonged together? What needed explaining? What had I perhaps misunderstood?

By the time I'd finished speaking, I'd done considerably more than provide AI with some raw material. I'd thought about the conversation again.

Bringing AI into the conversation

I asked AI to turn my spoken reflections into a document called Our Understanding, which I intended to share with the client. The purpose was to capture my understanding of their situation and give them an opportunity to confirm, challenge or correct it.

AI helped organise the material into a coherent structure (which I gave it in the verbal prompt), remove repetition and improve the flow.

Then I read the result. And, inevitably, I changed things. Some wording didn't quite capture what I meant. A few points needed more emphasis, while others needed less. I questioned whether certain conclusions were justified by what I'd actually heard.

I was making decisions again, drawing on my recollection of the conversation, my notes and my developing understanding of the client's situation. Another round of discernment.

What if I'd skipped all that?

Imagine I'd recorded the meeting, uploaded the transcript and asked AI to write Our Understanding.

I might have ended up with a perfectly good document. Possibly even a better one in certain respects, because a recording would have captured details my notes missed. But would I have understood the client's situation as well? I'm not sure. Perhaps the question is how much thinking I would have done between receiving the transcript and sending the document.

What I do know is that my process required me to engage with the material repeatedly, and each encounter demanded something slightly different.

During the meeting, I had to listen and select.

When I returned to the notes, I had to reconsider.

When I read them aloud, I had to organise and explain.

When I reviewed the AI output, I had to evaluate and decide.

Four encounters with the same conversation, each providing an opportunity to develop my thinking.

Of course, there's a risk here too. Every time we interpret something, we can introduce our own assumptions and biases. Repetition doesn't guarantee accuracy, and a polished narrative can reinforce a mistaken understanding just as easily as a sound one.

That's one reason the final document went back to the client for validation.

Discernment starts with us

We often talk about discernment in an AI-enabled world as the ability to evaluate what AI produces.

Check the facts. Challenge the assumptions. Look for errors. Don't accept the output simply because it's well written. That's important, but it's only part of the story.

Discernment also happens when we're deciding what information to gather, what to notice, what to leave out, how to interpret it and what questions to ask. It happens before we open the AI application, while we're working with it and after we've received the result.

And the more we practise discernment, the more opportunities we create to develop our own judgement. AI makes it increasingly easy to move from a conversation to a finished document with very little effort in between. Sometimes that's exactly what we need. Speed and efficiency have their place.

But I've come to appreciate the value of engaging with the same material several times, approaching it from different perspectives and allowing my understanding to develop along the way.

The output is the document. The other outcome is the understanding we've developed while creating it.I wonder how often, in our enthusiasm to make the process faster, we overlook that second outcome.