The cut, not the stone

A visit to the Diamond Museum in Amsterdam is a step into the world of craftmanship as well as a lesson on perceived value.

I ended up there by accident ten years ago, shortly after moving to the Netherlands. As a museum lover with plenty of time I decided to make a visit. Once inside, you see all sorts of diamonds, cutting machines and learn something I didn’t know: what gives a diamond its value is not the stone itself. It’s primarily how it is cut. The specificity of the cut reveals the artisan real skill so that two stones of the same size can have a totally different value based on the cut.

Now moving to the world of knowledge sharing, where two sources saying the same thing may create very different impact. Or have different audience size. Why is one followed and the other not? And why do we stick to certain authors or media?

It’s not just what they say, the content, but it’s how they say it. It’s their brand, their credibility.

So, while it’s true that AI is making access to knowledge almost commoditized, much of the current debate, focusing only on what is produced, is misplaced.

The disproportionate increase in how much content is produced and how easy it is to access is certainly doing something. It is increasing the amount of noise and making it hard to find uniqueness. As noise increases, it is becoming increasingly hard to distinguish what’s valuable and what’s not.

What does this mean for institutions whose product is knowledge? Many knowledge organisations, whose previous model was to spend 90% on research and 10% on dissemination, will have to rethink their model. The model was already struggling in the era of social media and shorter attention spans. It is entering an existential crisis now in the AI era.

The result is likely to lead, in my view, to higher costs of winning new audiences and higher loyalty to trusted channels. Kind of reverting what internet had created and made us used to.

The higher cost of winning new audiences is directly correlated to more voices competing for everyone’s attention. AI is only accelerating the trend, but the trend already existed.

However there’s a new element. Before the only issue was a lot of content competing for the audience’s attention. Now the new issue is that access itself is being gated.

AI is gating access to content, resulting in a general traffic decline. Reuters Institute reports data showing Google search traffic to more than 2,500 websites was down 33% globally from Nov 2024 to Nov 2025. Where specific data is available for the knowledge sector, data shows that 80% of culture organisations have seen a decline in search traffic in the UK. This kind of decline affects more educational content rather than transactional one.

Increased loyalty is an interesting new element. Although correlated to the higher cost of winning new audiences, it’s a new potential feature of the AI era. Here’s what makes it plausible: largely doubting the origin of content, our natural tendency as audience will be to default to the sources we trust.

For knowledge organisations, the first effect is the increased importance of investing in trust signals and curating each single interaction. As interactions lower and their cost increase, there is more to lose from interactions that are less than perfect or off brand.

Following the first, the second aspect is the importance of a clear trusted presence, a brand. Branding is a quality signal in itself. When well executed is signal that every piece of content, across every platform, is worth paying attention to.

In the end it’s the cut and not the stone that drives value, as the lesson from diamonds helps remember. AI has not changed the rules of the game. But it has increased the cost of ignoring them.

‍