Why is AI losing the communications and PR war?

In May 2026, the former Google chief executive Eric Schmidt stood up at the University of Arizona to tell graduating students about the AI-driven future waiting for them.

They booed him.

The class of 2026, as Slate put it, is sick of hearing about AI, and a petition to remove him as speaker had already gathered more than 1,200 signatures. Across American campuses that graduation season, a pattern held: a technology leader takes the stage to promise a transformed world, and the people that world is being promised to walk out, or jeer, or simply stop listening.

That image is the whole story in miniature. The argument for what is purportedly the most capable technology of our lifetimes – very visibly promoted by some of the richest and most powerful people alive – is steadily being lost with the public. In May 2026, an Economist/YouGov poll found 71% of Americans think AI is developing too fast, against 2% who say too slow. Opposition to the data centres that are the physical backbone of the whole enterprise has climbed steeply as the build-out accelerates (with a ban proposed in my native Scotland due to the hostile sentiment).

If you believe those who espouse its potential and the virtues, then this growing hostility towards AI becomes the biggest communications disaster of our age. What on earth has happened?

To me, it feels all too clear: AI's leading voices are losing the communications war because they are fighting a publicity war instead. They are spending on the message and neglecting the fundamentals. And the fundamentals of good communication were and are never about persuasion from a stage. They are about understanding people, meeting real needs, and showing tangible good rather than announcing it.

Buying the microphone instead of earning the room

Start with the clearest tell. In April 2026, OpenAI bought TBPN, a daily tech talk show, for a sum the Financial Times put in the low hundreds of millions. It was the company's first media acquisition, and the revealing detail is where it landed internally: not in a content or media division, but inside the Strategy organisation, reporting to the chief global affairs officer. The New York Times' Silicon Valley correspondent called it plainly, a marketing expense, arriving precisely as public scepticism was rising.

Think about what that decision says. Faced with a collapse in public trust, the response was to acquire a friendly platform: to own more of the conversation rather than change what is being said in it. That is the instinct of people who believe their trust problem is a distribution problem, that if they could only get the message in front of enough people, enough times, the doubt would clear. It is a category error. You cannot buy your way to being believed. Trust is not share of voice.

No message survives contact with a higher bill

The deeper failure is that AI's champions are talking while the public is feeling. And what people are feeling is the cost, in their own lives, directly.

Take the one that will reach almost everyone reading this. The AI build-out has created a global shortage of memory chips, because manufacturers are diverting production to the high-margin memory that AI data centres need. AI is on track to consume around 70% of the world's memory output in 2026, up from 20 to 30% a few years ago. The result lands in your pocket. Apple has confirmed price rises across its products, with Tim Cook citing the memory shortage as the reason. Your next phone and laptop cost more so that the data centres can be fed.

Nobody communicated that trade-off in advance, or even hinted at it – never mind to seek understanding or agreement. It simply arrived, with an 'aw shucks' attitude from those responsible, if they even bothered to acknowledge it.

Then the data centres themselves. In the UK, campaigners held coordinated days of action in February 2026 against developments in Buckinghamshire and Essex, on exactly the grounds you would expect: water, power, and household bills. The regulator Ofgem has disclosed around 140 data centres seeking grid connections, and the largest single site planned would reportedly draw more electricity than double all the households in Wales. The friction has only grown since: in London, a £500m data centre proposed on Brick Lane is being fought by residents who argue the land should be housing, while a £2bn project in Essex has stalled in the grid and net-zero backlog. In the United States the resistance is further along still, with more than 75 projects worth around $130 billion blocked or delayed in the first months of 2026 and dozens of local bans. This month New York went further again, becoming the first US state to impose a statewide moratorium on new data centres, its governor citing residents' utility bills, and authorities around the world are now restricting them.

You cannot spin the prospect of an electricity bill. When the abstract promise of intelligence collides with the concrete experience of a costlier phone, a strained grid and a data centre proposed next to the school, the promise loses every time, because one of them is real to the person living it – and the other is a flashy slide in a keynote.

Telling us it is good, from inside a bubble

Which brings me to the register, the actual posture of what we've seen and heard over the past couple of years. Overwhelmingly, the public case for AI is made by a very small number of extraordinarily wealthy people, telling everyone else that this is good for them. It is communication in the imperative and the abstract: the future is inevitable, the gains are civilisational, trust us.

I have written before about the two registers AI is usually sold in, hype and doom, and how both spend public trust rather than build it. But the graduation booing points at something more basic than register. It is the direction of travel. The message flows downhill, from people insulated from the costs to the people who will carry them, and it asks for belief without offering evidence the audience can touch. That is not how trust has ever been built. It is the posture of telling, when the only thing that persuades is showing.

And here is the part that should sting, because the good is real – and there are stories to be told. AI is already doing tangible good: in critical sectors such as health, in diagnostics, in accessibility, and now increasingly in the unglamorous back offices of organisations that serve people who genuinely need serving. This work exists. It is simply not what AI's loudest voices choose to communicate, because it is small, specific, human and slow, and they have trained themselves to speak in the language of the epochal instead. The most persuasive case for AI is being left on the floor while the money goes on a podcast.

What the fundamentals actually look like

I do not say any of this as a sceptic. I spend my working life helping organisations communicate, and I have watched AI do genuinely useful things up close. I say it as someone who thinks its leading advocates are getting the easy part wrong.

The easy part is the fundamentals, and good communicators know them cold. You start from the audience, not the message: what do these specific people need, fear and value? You build with them in mind rather than at them. And you show rather than tell, because one demonstrated benefit in a real life is worth more than a thousand promises about humanity. When I founded Covid Aid in 2021, none of it worked because we announced that we were helpful. It worked because it met a real, specific need, and people could see it meeting theirs. That is communication doing its actual job, and it is the opposite of buying a microphone.

AI's leading voices have almost unlimited resources and some of the most consequential good-news stories of the decade to draw on, and they are losing the communications war anyway. Not because the technology is weak, but because they have confused communication with promotion. They are telling a sceptical public what is good for them, from a stage, while the bill lands on the kitchen table. The way back is not a cleverer campaign. It is the oldest discipline in the craft: understand people, build for their real needs, and show them the good instead of insisting on it. Until they relearn that, the booing will get louder, and they will have earned it.

Applied / Comms With AI documents what actually works in AI for communicators, and what does not. Explore the free resource at Comms With AI, and the consultancy behind it at Faur.