Overclaiming Is a Silicon Valley Tradition

I was there at the beginning of the commercial internet. The late 1990s. The promise was simple and seductive. For the first time in the history of advertising, you would know exactly what worked. Every click tracked. Every conversion attributed. John Wanamaker's century-old lament, that half of all advertising was wasted and he just didn't know which half, was finally going to be solved.

I didn't believe it then. The attribution problem was structural, not technical. Knowing someone clicked an ad tells you almost nothing about why they bought. The industry was measuring what was measurable and calling it proof.

I said so at the time. Not in writing. In rooms.

It didn't matter. The underlying technology was real. The market believed the specific claims because it wanted to. And the correction took a long time to arrive. By the time it did, digital advertising was so embedded in the media mix that the overclaiming had cost the industry a reputational price but not a business price. Adoption was already locked in.

The AI industry is falling into the same trap.

Two claims being made with certainty the evidence does not yet support. AI as existential safety risk. AI as the great job killer. The technology is real. The specific claims are not. And just as the early internet executives believed their own attribution story, I think the AI industry believes its own narrative. That makes it more dangerous, not less. Cynical overclaiming gets corrected faster. True believers are harder to reach.

Here is where the parallel breaks down in a way that should concern the industry.

The internet had decades for the overclaiming to compound before the correction arrived. Everything about this AI era is moving faster. The backlash that took thirty years on the internet is arriving in real time. Communities are already protesting data centers. Public sentiment is already negative. The correction is happening before adoption is locked in.

Digital advertising survived its overclaiming because the market caught up too late to matter commercially. AI may not get that runway. The claims are more extreme. The timeline is shorter. The blowback is arriving before the behavior is entrenched.

If the industry doesn't correct the overstatement quickly it may pay both prices. The reputational price is already accumulating. The business price may be avoidable.

The playbook is the same. The timing is different. That changes everything.