Cost Signalling in the Era of AI
The mechanism connecting the analogue renaissance to the human premium.
I have written two pieces recently that generated more sustained conversation than anything else I have published. The first was about the analogue renaissance: the cultural migration toward things that feel irreducibly human. The second was about the human premium (pt.1, pt.2): the thesis that human contribution appreciates as AI commoditises everything formulaic.
Both kept pointing at each other. Readers noticed it too. Several people asked, in various ways, the same question: are these two ideas connected?
They are connected. And the mechanism that connects them has a name. It is called cost signalling.
The signalling problem
There is a concept in economics and evolutionary biology called signalling theory. The central insight is simple: a signal is credible only when it is costly to fake.
A peacock’s tail signals genetic fitness because growing and maintaining it is metabolically expensive (as well as making it easier prey - the ultimate cost). A university degree signals competence because it takes years and cannot be purchased overnight. A Michelin-starred restaurant signals quality because the process of earning and maintaining stars is gruelling and public. In each case, the signal works precisely because it cannot be cheaply imitated.
Writing used to work the same way. A deeply reported piece in a serious publication carried implicit cost signals. The research, the editorial oversight, the time investment were all baked into the format itself. You did not need to be told that a 10,000-word investigation required months of work. The depth was the signal.
AI has broken this. The format can now be replicated without the investment. A generated article can have the same structure, the same length, the same apparent depth as something that took weeks to produce. The implicit cost signal has been neutralised. And when implicit signals fail, a trust vacuum opens.
That vacuum is what both the analogue renaissance and the human premium are responding to, from different directions.
Demand meets supply
The analogue renaissance, as I described it, is a demand-side phenomenon. People are gravitating toward things where the cost of production is visible and cannot be faked. The reason a handmade object, a live performance, or a face-to-face conversation carries weight right now is not nostalgia. It is that these things function as credible signals in an environment where most output has lost its signalling power.
The human premium is the supply-side counterpart. When AI handles everything formulaic, the remaining human contributions, the ones involving judgment, taste, accumulated context, and the capacity to be genuinely wrong, become the scarce inputs. They are expensive to develop, impossible to shortcut, and difficult to verify through the output alone.
Cost signalling is the bridge. It is the mechanism by which supply becomes legible to demand. Without it, the person who spent three weeks thinking through a problem and the person who spent thirty seconds prompting a model produce outputs that look the same. The market cannot always tell them apart. Cost signalling is what makes the difference visible.
Why quality is no longer enough
This is the part that is counterintuitive and worth sitting with.
The argument is not that expensive things are better. Plenty of expensive things are terrible, and AI can produce genuinely excellent output. The argument is that when quality becomes un-differentiable at the point of consumption, cost becomes the only remaining signal of provenance.
Think about it in terms of food. There was a period when processed food and whole food were visually indistinguishable on a supermarket shelf. The product inside the packaging might be entirely different, but the consumer had no way to know. What emerged was labelling. Ingredients lists. Provenance markers. Organic certifications. Not because consumers studied every label in detail, but because the existence of the label signalled accountability. It said: we are willing to be scrutinised. That willingness was itself the signal.
Content is reaching the same inflection point. The question is not whether AI-generated content can be good. It obviously can. The question is how a reader, client, or audience member distinguishes good content that emerged from genuine thought from good content that emerged from a well-crafted prompt. At the point of consumption, they often cannot. The output, save for a few tells to a well trained eye, is identical. The provenance is invisible.
Making cost visible
If cost signalling is the mechanism, the practical question becomes: how do you make cost visible without it becoming performative?
I have been thinking about a few approaches.
The first is what I think of as an ingredients list for content. A structural element, not a footnote, that states plainly what went into a piece: the duration of conversation it was drawn from, the number of editing rounds, the sources consulted, whether and how AI was involved. The logic is borrowed directly from food labelling. It works not because every reader will scrutinise the list, but because the act of disclosure signals that the cost is real and the author is willing to have it examined.
The key insight is that the ingredients list only works when the investment is genuine. If you label a piece “90 minutes of recorded conversation, three editing rounds, two domain experts consulted” and the article reads like a shallow summary, the label actively damages credibility. The cost must be real for the signal to function. This is what separates cost signalling from marketing. Marketing makes claims. Cost signalling invites verification.
Since writing this I have done a bit of research and found several organisations working on a similar idea for certifying human-generated creative work, most representing something close to an organic labelling-esque solution.
The second approach is format as implicit signal. A polished dialogue between two named participants carries embedded provenance that a solo article does not. The format itself communicates that two people sat down, exchanged ideas, disagreed, and built on each other’s thinking. That texture, the digressions, the self-corrections, the moments where someone changes their mind, is structurally difficult to fabricate. The format is the evidence.
There is also a distribution advantage here. When content is structured as a conversation between two people, both have an incentive to share it. Their credibility is attached to the output. The organic reach multiplies without additional promotion, because the cost of collaboration, coordinating schedules, having a real conversation, editing together, functions as its own signal.
The third is what might be called demonstrated thinking in public. Not thought leadership, which has become another way of saying “opinions presented as authority.” Demonstrated thinking is messier. It includes the reasoning, the uncertainty, the places where the argument is still forming (like here!). It is expensive to produce because it requires actually thinking, not just presenting conclusions. And it is expensive to fake because the texture of genuine intellectual process, the specific wrong turns and self-corrections, is difficult to simulate convincingly.
The unified picture
Here is how the three ideas fit together, stated once and cleanly:
The analogue renaissance identifies a market shift in demand. People want things that feel real, not because they are sentimental, but because real has become scarce.
The human premium identifies a corresponding shift in supply. The human contributions that remain after AI handles the routine are precisely the ones that are costly to develop and impossible to automate.
Cost signalling is the mechanism that allows supply to reach demand. It is how the genuine article makes itself legible in a market where imitations are free and abundant.
For anyone producing work in the ideas space, the implication is direct. The work still matters. It always will. But in a landscape where the output of genuine effort and the output of a prompt can look identical, the ability to make the effort visible is becoming the decisive competitive factor.
The content market is not dying. It is bifurcating. On one side, an ocean of zero-cost, zero-provenance output that audiences increasingly scroll past. On the other, a smaller body of work that carries visible evidence of human investment. The gap between those two categories is where credibility lives now.
Cost signalling is how you get to the right side of that gap.
If this piece added something to your week, please consider subscribing :)