As AI makes everything online easier to doubt, physical experience becomes marketing’s proof of humanity.
For years, the bottleneck in advertising was production.
Could you make enough good content? Could you make it quickly enough? Could you create enough films, images, copy, edits, variants and social assets to feed the machine?
Generative AI has changed that equation. There is no longer scarcity of creation. The tools are cheap, fast and widely available. Adobe’s GenStudio is explicitly built around generating on-brand copy and creative asset variations at scale, while Firefly Services brings generative AI into creative workflows and production pipelines. TikTok’s Symphony Creative Studio points in the same direction, giving advertisers AI tools to create and adapt content for the platform in minutes. Online content production is no longer a specialist capability, it has been democratised.
The problem of content oversupply
The internet is entering a phase where more content can be made than people can meaningfully consume, platforms can reliably rank, or brands can confidently measure. A recent study from Stanford, Imperial College London and the Internet Archive found that by mid-2025 roughly 35% of newly published websites were classified as AI generated or AI assisted, up from almost zero before ChatGPT’s launch in late 2022. The study does not prove every fear about AI content, but it does show the scale of the shift.
So the problem is no longer whether content can be made. The problem is whether that content is seen and trusted by humans and, in turn, can drive real commercial value.
That is where digital advertising starts to get uncomfortable. If AI can generate the content, automation can also consume it, scrape it, amplify it, interact with it and distort the signals we use to judge success. Impressions, reach, views, clicks and engagement were already imperfect measures. In a synthetic media environment, they become weaker still.
This is not theoretical. Thales’ 2026 Bad Bot Report says bots accounted for 53% of all web traffic in 2025, with bad bots alone accounting for 40%. HUMAN Security’s 2026 State of AI Traffic report points in the same direction, reporting that automated traffic grew eight times faster than human traffic year on year, AI-driven traffic grew 187% across 2025, and traffic from AI agents and agentic browsers grew 7,851%.
Machine traffic is accelerating fast. This does not mean every impression is fake. It does not mean every view is worthless. But it does mean the evidential value of those metrics declines even further.
The dying internet can’t offer human verification
The “dead internet theory” started life around 2021 with the idea that much of the web was no longer real people, but bots and machine-generated content talking to each other (see The Atlantic article for more). For a while, it sat comfortably in the bucket of internet folklore, easy to dismiss. What makes it interesting now is that the numbers have started to catch up with the paranoia, the conspiracy theory is quietly becoming a measurable commercial fact. Digital media has a verification problem. Not just “was the content real?” but “was the audience real?”
If brands want confidence that they are reaching actual people, they need moments that cannot be faked. Environments where participation is intentional, and interaction can be observed and properly measured.
A live experience, retail activation, event or immersive brand world can give you something the internet cannot offer: verified human presence.
You know a person was there because they entered the space. You know they chose to engage because they spent time, asked questions, interacted with the product, spoke to someone or moved through a designed journey. On the internet, a view tells you something appeared on a screen. A like tells you someone, or something, clicked. A physical experience tells you a person showed up.
For evidence of how much that presence is worth, look no further than Taylor Swift. The Eras Tour became one of the highest-grossing tours in history and was credited with measurable boosts to local economies wherever it landed. “Swiftonomics” entered the business pages because a physical, unfakeable event moved money in a way no volume of online impressions could.
Don’t lose all hope on digital, but rethink it
Digital still matters. It gives us reach, targeting, CRM and connected, measurable journeys. But blind faith in digital metrics should be over. The next phase of brand experience needs to connect digital systems to the real world.
That means designing experiences where data is not just collected from passive exposure, but from meaningful participation. Dwell time matters more when you know someone physically chose to stay. A qualified conversation matters more than a click you can’t trust. A configured product, booked appointment then an in-person follow up tells you far more than an impression ever could.
The verification bottleneck
The production bottleneck has gone. The verification bottleneck has arrived.
That is the real marketing problem. Digital platforms can still count activity, but activity is no longer the same as human attention. In a synthetic internet, brands need evidence, not just numbers. Physical experience becomes the closest thing marketing has to a CAPTCHA: proof that a real person was there, chose to engage, and spent time with the brand.
AI makes content easier to produce, but easier to doubt. Physical experience is harder to produce, but harder to dismiss. In that environment, physical experience is not a nice-to-have layer on top of a media plan. It becomes one of the few places where brands can still evidently prove human interaction happened.
Real people. Real interactions. Real value. That is where marketing spend starts to matter again.
This first appeared in Performance Marketing World, written by Simon Levitt, Global Creative Technology Director from our London studio.
Behind the piece
More like this
July 28, 2026
The Transformation Economy – why experiential marketing should start targeting progress
