Your AI pilot didn't stall on the technology — it stalled on the infighting

If your AI pilot quietly went nowhere, it almost certainly did not die on the technology. It died in the meeting, where finance, marketing, IT and the C-suite each spoke their own language, defended their own budgets, and agreed on an outcome they had each privately defined differently. In the corporate world we don't call it infighting. We call it politics. It is the same thing, and it is where AI strategies go to stall.

I have watched this my whole life. When I was eleven, at the London Toy Fair, I walked past the Sega stand where a room full of buyers from the big retailers were watching adults fail to get a console working. I stepped up, unplugged something, plugged something else in, sat down and started playing Sonic. They sold out within half an hour, because their target market had just shown them it worked. The technology was never the problem. The room was.

Why do AI pilots really stall?

Because the friction is human, not technical. The data is blunt about it. RAND (the RAND Corporation, the US non-profit research institute that advises governments and the military on hard policy questions) studied why these projects fail and found that more than 80% do, roughly twice the failure rate of IT projects that do not involve AI. The leading root cause was not the model. It was that business leaders and technical teams understood and described the problem differently, so each side built towards a different finish line. MIT puts the headline figure even higher: around 95% of generative AI pilots produce no measurable financial return (MIT, 2025). The pilot works. Then it meets the org chart.

I once watched a business stall a solar-panel investment that would have paid for itself in eight months, purely because the cost came from the capital budget while the savings landed in the operational budget, and two different people controlled the two budgets. Logical action, blocked by silos. AI lands in exactly the same too-hard basket. Gartner (the technology research and advisory firm) predicted that at least 30% of generative AI projects would be abandoned after the proof-of-concept stage by the end of 2025, citing escalating costs and unclear business value (Gartner, 2024). Most of those projects worked in the demo. They died in the negotiation that followed.

What does infighting actually cost?

The opportunity, usually. In my family's toy business in 1995, I wanted to build a proper website: games for the kids, a card on their birthday, offers matched to their age. We could have been the Amazon of toys before Amazon. The board could not see it, so we built something plain instead. A year later, when a colleague was selling pay-as-you-go mobiles, a board member asked, in all seriousness, "why would children want a mobile phone?" By the time the infighting resolved, the lion's share of every one of those opportunities was gone.

That is what dissonance costs. While you argue about strategy, an aligned competitor is already testing, measuring and iterating. Sharyn Leaver, chief research officer at Forrester (a global market-research firm whose analysts study how companies adopt technology), summed up the pattern bluntly in 2026: AI urgency is at an all-time high, yet too many businesses are paralysed by a lack of understanding and siloed adoption (Forrester, 2026). The friction of infighting makes you slow, and slow is how you end up playing catch-up on a position that sat at the front of the market when the argument started.

How do you break the infighting?

Infighting is organisational dissonance, and the answer is coherence. Not everyone agreeing on everything, but everyone aligned on the same problem, the same outcome, and a shared language they can use to discuss the work. Four moves:

1. Agree the problem before the solution. Most rooms are arguing about answers to a question they never defined. Name the problem first, in one sentence everyone accepts.
2. Agree a single outcome. Keep chunking up (a coaching technique of asking "what is that in service of?" until you reach the broader goal underneath), past "marketing wants a tool" to "we want to serve the customer better", until there is one outcome everyone shares.
3. Translate out of jargon. Acronyms are how experts defend turf. Force plain language and the value conflicts surface where you can resolve them.
4. Bring the evidence, then decide. Lay out the problem, the outcome, the cost of inaction, and the data, and then it is just rational people making up their minds, not egos defending ground.

When the infighting stops, the thing leaders always notice is energy. The grey, apathetic, zombie version of the organisation lifts. People stop fighting to be right and start building together. That is coherence, and it is the precondition for AI working at all. The bottleneck was never the technology. It was the dissonance in the room.

About the author

Thomas Green

British technology futurist, AI keynote speaker and advisor. Thirty years across enterprise technology and AI strategy, helping leaders navigate the future of work. The futurist who died.