Doing less
There’s an old cliché; are we human beings, or human doings?
AI forces the question into the open.
Navigating spreadsheets? Typing boilerplate code? Summarising meeting notes?
We now know these are not uniquely human tasks.
But we probably knew that before too.
Nothing else could do them reliably, so we did. But did they light us up? Did they give us a sense of immanence, of presence, of vitality that made us glad to be be alive?
It’s important to acknowledge, thought that those tasks weren’t entirely without value.
Going deep into spreadsheets could help us understand the ground truth of our business. Writing boilerplate code could give an appreciation of software design decisions. Summarising meeting notes could give a chance to digest, reflect and share not just information but insight.
But these were a means to an end. We did these tasks in service of a higher purpose - seeking truth, learning, sharing. And if we can reach the same ends faster, we will.
But what happens if and when AI starts moving up the stack? Not just navigating spreadsheets, but managing portfolios and allocating capital on global levels? Not just writing code, but researching, prototyping, experimenting and productionising entire applications and businesses? Not just summarising meeting notes, but managing communication across millions of agents?
These might still seem like sci-fi, but ChatGPT 3.5 only launched in November 2022. It hasn’t even been 5 years, and the landscape of software is unrecognisable, and is likely to change even faster from here on out - to stay nothing of wider knowledge economy jobs.
So where does this leave us? And what can we do in the face of such rapid and impactful change?
I think the real question isn't how we do more with AI. It's how we learn to do less — and be more present in what remains.
Join me for my live later today where I’ll discuss this further.
The distinction you draw between the task and what the task was in service of is the right one. Spreadsheets as a means to understanding ground truth. Boilerplate code as a means to developing design intuition. The task was never the point. The formation that happened while doing it was.
What I find myself adding is the harder question underneath. Doing less sounds like relief until you sit with what the tasks were doing to you while you did them. The spreadsheet that built your understanding of the business. The code that gave you the feel for why a decision was made. Strip those out and you reach the insight faster. You also skip the development that would have happened in the slower path.
The question is not whether we can reach the same ends faster. It is whether the ends we reach via the faster path are actually the same ends. Or whether something that looked like a means was quietly also the point.
Doing less with AI is only a gift if you know what you were actually building while you were doing more. Most people have not had to think about that yet.