Increments

There is a quiet power to increments.

Almost unnoticeable individually, they gather together until, suddenly, everything is different. Compound interest is about increments. So is nigredo, the initial stage of alchemy that starts transformation quietly and unnoticeably at first. It does not need permission, nor can it be invoked; it is a natural part of any system. 

I'm writing this on a train going from Derby to London. It is a good, if unfortunate, example of increments. Those who use it regularly probably don't notice its quiet deterioration in all the small areas - the wifi, the overcrowding, the regular, minor but inconvenient unreliability, and the steadily increasing cost. The frequent passengers adapt. I'm fortunate in that I use it only infrequently, and never under time duress, but it does mean I notice the aggregation of increments. Over time, it has become something to avoid if at all possible.

What really interests me though, is the nature of increments when it comes to our work lives as AI finds it's way into the cracks. It's tempting to be alert to what it might mean on the constitution of our jobs, but that misses the point. Jobs won't change incrementally, but the structure of work will.

I suspect the first we will know about the effect on our jobs is either when they disappear, or we find ourselves unqualified for them; when we find ourselves credentialed for them, but not competent. When we find technology can do the processes, but not the relationships that connect them to other jobs because we have become so efficient and so productive we do not understand the ecosystem in which they exist. 

People who live and work in complex systems, like diplomats, or investment and intelligence analysts hone their skills in the spaces beyond where the data ends. Their mastery exists in the half tones of nuance, tone, habit and other qualities that are difficult to see when they are working at maximum capacity.

We don't notice increments when we're busy; we need space and the company of strangers who see things differently to us to notice them before they turn up, in a gang, and surprise us. As Mike, in Hemingway's "The Sun also Rises" observed, bankruptcy happens gradually, then suddenly. Every company, and every job, remains individually unique even as they become increasingly homogenous. One bank, one utility company, one phone company is only incrementally different to another. They remain focused on each other as competition even as the system they are part of becomes obsolete.

The same is true of the jobs within them; one account manager, one accountant, one manager posts why they are different on LinkedIn, even as technology, increment by increment, makes their current skills obsolete. Not them as individuals, their skills. I suspect it's like language. I spent five years working in Switzerland, in two languages; my insufficient German, and my hosts good, but insufficient English. We thought we understood each other, and were often too proud to admit where we did not. What happened in the space of incremental incomprehension between us, in retrospect, makes me cringe.

The increments that gather in professional work are following the same pattern as that train journey, small changes accumulating until the experience becomes transformed. What’s fading isn’t disappearing overnight. It’s more like erosion. The hours spent gathering information, filling templates, conducting routine research; the scaffolding tasks that once filled mornings and afternoons. They’re shrinking, becoming almost incidental, like a conversation partner who gradually speaks less until you realize you’re doing all the talking.

What’s expanding instead lives in those spaces beyond where the data ends. The difficult judgments, and pattern recognition across complexity. The strategic thinking that exists in half-tones of nuance and context. The work that, like those diplomats and analysts who operate in territory too subtle for algorithms to map.

Something new is emerging in the spaces, a capability that barely existed two years ago. The art of working alongside AI: knowing what to ask, what to trust, and when to override. The critical skill of knowing when not to use it. It’s becoming as fundamental as literacy, yet we’re not teaching it. Universities offer it in fewer than one program in twenty. Faculty themselves are still learning, still proud, still occasionally too uncertain to admit the gaps.

What troubles me most is that we’re teaching people to use these tools, but not to question them. Not to recognise when AI is confidently wrong, or when human judgment should prevail. We’re optimising for adoption while the increments of incomprehension accumulate between us and the technology, just like those conversations in insufficient German and insufficient English, where pride masked the growing space of misunderstanding.

The professionals who will thrive aren’t resisting this or rushing toward it. They’re the ones comfortable holding contradictions, domain experts who embrace technology while remaining fundamentally skeptical of it. People who can leverage AI’s efficiency without losing sight of the ecosystem, the relationships, the context that makes work meaningful rather than merely processable. That hybrid capability, technical enough to collaborate, wise enough to question, experienced enough to see the increments gathering, is becoming the scarcest resource in professional work.

Not because it’s revolutionary, but because like compound interest, it requires patient attention to accumulation that most of us, busy and focused on competition, simply don’t notice until we find ourselves credentialed but incompetent, efficient but obsolete.The bankruptcy of relevance, as Hemingway understood, happens gradually. Then suddenly.​​​​​​​​​​​​​​​​

That’s why The Athanor exists. To provide the space to notice, and to learn, and to make mistakes before those mistakes become important.