Seeing is no longer believing

If the technology worked, as you read this I am somewhere in New Mexico on a bicycle and a tent; no laptop, no meetings, and hopefully no snakes.

This edition is about being discombobulated.

In 2023, Merriam-Webster made authentic its word of the year. Lookups had kept increasing, driven by artificial intelligence and deepfakes. Deepfake made the same year's list.

Two years later, in December 2025, they chose slop: digital content of low quality, produced in quantity by artificial intelligence.

I wrote a guide in July called AI customer stories, without the slop. It had not occurred to me that I was quoting Merriam-Webster.

01  You genuinely cannot tell

It's not just my age; I think this applies to all of us.

We scroll past images, videos, reviews and posts every day without being sure which of them involved a human being, and to what extent. Looking closely does not help.

The technology has evolved to the point where we cannot tell, so we cannot trust.

That is a new condition for a species that has always used its eyes to decide what to believe. Seeing used to be believing. This change happened in about thirty months, and there is no putting it back.

We now all believe everything a little less. Politics and misinformation do not help either. That less-trusting habit does not stay on Instagram; it comes to work on Monday, and it reads your case study.

The neutral buyers, in that data, did not become believers. They became sceptics.

02  The good half of this

I've spoken about this before; there are two sides to the change, and one I love.

The barriers to creating art and B2B content (sometimes the same thing) have collapsed. Anyone with a real idea, some taste and the patience to iterate can now produce work that five years ago needed a budget, a studio and months of somebody else's time. That is not a small thing. It's the widest and most exciting redistribution of creative capability in my working life, and I am glad to have seen it.

The flipside arrived in the same box. The same function that allows an individual with an idea to make something wonderful with almost all constraints removed lets everybody else generate infinite, adequate, nothing. Slop.

So we get both: an explosion of fast, excellent creativity, but floating in a sea of vastly more sludge than ever, with the reader/viewer unable to sort one from the other.

03  Which makes our job more valuable, not less

A natural conclusion could be that customer stories get devalued along with everything else. I think the opposite is true if we do things correctly.

People trust people like them. That has been true for as long as humans have been able to communicate.

Edelman's barometer keeps finding trust needs proximity: my employer, my colleagues, the people I can see. The B2B version of that shows up in the buying data, where half of buyers went and spoke to an actual peer.

Vendor collateral came last.

A customer story is the only asset our profession produces that provides proximity needed for trust, at a distance. A named human being, at a real company, describing what happened to them, conveying their story to somebody who will never meet them.

It only works if the reader believes the human is real.

In a slop economy that is worth more, not less. We have always done the hard part; the real person, the real quote, the sign-off. What we have never done is show it.

04  The law has noticed too

This August, the EU's AI Act transparency rules came into force, and the major model makers now weave invisible marks into generated text. Most of what I have read about this is wrong, so here is the version that matters for our profession.

The marks travel with the text. There are already articles about how to strip them out; futile, and not the point. If your content strategy depends on hiding how the work was made, the watermark is not your biggest problem.

Why did the EU act? Deception at scale: deepfakes, fake influencers, fabricated customer testimonials. Good. I am delighted about this.

But be clear about what a mark means. It means a model touched the text; not that the model made it up. Ask a model to tidy a paragraph you wrote yourself and it will come back marked.

What the rule is actually aimed at is fabricated provenance: a person who does not exist, or a real person who never said the words and never signed them off. A watermark cannot detect that either. Only structure and attribution can.

Nobody was ever asking who typed it. They were asking whether the customer meant it.

05  The story's existence is no longer the proof

For twenty-five years the story was the evidence. You produced it, you put it on the website, and its very existence was the proof. We trusted, and normally were right to do so.

So the story now has to carry its own evidence. Not more prose: structure: a named person at a named organisation; an approval recorded as an event with a date rather than a vague assurance; a permanent address that survives the next site redesign; and a machine-readable layer, because the first reader is now often not a person, and it will not guess in your favour.

None of that is editorial or design work; it's a different structural strategy for customer stories in the age of distrust.

This, as ever, is why Proofpoints publishes a standard rather than a promise, and why that standard says plainly what it does not cover. "Trust us" stopped being an available position somewhere around 2024.

06  What comes next

This is the sixth issue of Proof Positive.

Next issue: showing my working. The thing I have spent eighteen months building goes generally available, so I will do what this issue asks of everybody else; publish the story, then show the working underneath it, including mistakes I made.

Subscribe if you have not. Forward to one person whose stories deserve better evidence than they currently carry.

§

Robin Hamilton is the founder of Proofpoints, the customer storytelling platform launching in 2026. Previously, cofounder and CEO of inevidence. Proof Positive is published fortnightly.

Two weeks from now, in your inbox. Fortnightly, no spam, easy unsubscribe.