When AI Adoption Becomes Theater: Inside Broken Enterprise Rollouts

The Strategy Gap Nobody Wants to Admit

When consultant Dan Boyles sat down with the C-suite of an oil and gas company and asked why they were adopting AI, he got three different answers from three different people. The CEO wanted to keep up with competitors. The head of sales wanted more revenue. Marketing wanted to cut contractor costs.

None of those are AI strategies. They’re wishes.

This kind of misalignment at the top doesn’t just slow things down — it poisons the whole rollout. Teams receive vague mandates, pick tools at random, and measure success by vibes. The ROI never materializes, and everyone quietly blames the technology.

The real culprit is the absence of a single, honest answer to a simple question: What are we actually trying to fix?

When Adoption Becomes a Metric

KPMG built a dashboard to track whether US employees hit a 75% AI usage target. Accenture reportedly tied promotions to “regular adoption of AI tooling.” These are real policies at real firms.

On the surface, it looks like commitment. Underneath, it’s measurement theater.

Tracking usage doesn’t track value. An employee who opens an AI tool twelve times a day to generate summaries they immediately discard is hitting their target. An employee who uses it once a week to cut a three-day analysis down to four hours is the one actually moving the needle.

Confusing activity with impact is an old management mistake. AI just makes it more expensive.

The Culture Problem Nobody Budgets For

Here’s the uncomfortable truth: AI doesn’t fix broken cultures. It accelerates them.

Caroline Rawlinson, CEO of Culture Amp, puts it plainly — drop AI on top of a fragmented or fear-based organization and you get, at best, a painfully slow rollout. At worst, a very expensive pile of nothing.

And the numbers back this up. Nine in ten HR professionals expect to increase their use of generative AI. But a third of those same professionals say no one currently owns AI strategy at their companies.

That’s not a technology problem. That’s a leadership vacuum dressed up in a software subscription.

Change Done *To* Workers, Not *With* Them

The UK government wants AI to “rewire” the state and boost public sector efficiency. Civil servants, for their part, are broadly open to the idea. But research from the FDA union found that fewer than a third had been consulted on how the rollout would actually work.

The union’s general secretary described it simply:

“Change is being done to workers, not with them.”

This pattern repeats across industries. Mandates flow downward. Training is either absent or checkbox-deep. The people who actually understand the workflows — the ones who would know where AI genuinely helps — are the last ones asked.

Unsurprisingly, adoption stays shallow and productivity gains stay theoretical.

What Good Actually Looks Like

The oil and gas story has a second act. After the C-suite confusion, the company’s president eventually got honest: he wanted to increase operating earnings because he planned to sell the business within a few years.

That one clear motivation changed everything. Boyles’ team could map actual processes, find real bottlenecks, and identify where AI would create measurable value — rather than impressive slide decks.

Specificity is the unlock. Not “we’re embracing AI” but “we need to reduce time-to-quote by 40% in the sales cycle, and here’s the workflow where that happens.”

The Ingredients That Actually Matter

A senior consultant at a large firm — speaking anonymously — described a more grounded approach: everyone gets access to two core tools, specialists can request more, and nobody gets access without mandatory training covering AI ethics, bias risks, and the very real tendency of these tools to hallucinate confidently.

A few things stand out in that model worth noting:

Ownership is clear. Someone is accountable for the strategy, not just the subscription.

Access is earned, not assumed. Tools are matched to actual job demands, not distributed like branded tote bags.

Human variables are acknowledged. Generational differences in confidence, varying comfort levels, different learning curves — these aren’t edge cases. They’re the rollout.

The Takeaway

AI adoption theater is easy to spot in hindsight and surprisingly easy to stumble into in real time. The pressure to signal progress is real. The fear of falling behind competitors is real. But tools deployed without strategy, culture, or genuine employee buy-in don’t transform organizations — they just add noise.

The firms that will actually benefit aren’t the ones with the highest usage dashboards. They’re the ones who asked the boring, unglamorous question first: What specific problem are we solving, and for whom?

Everything else is just a very expensive performance.