Corporate executives are forcing artificial intelligence tools onto their workforces with zero clear strategy, and I am not surprised to see it blowing up in their faces. A rush to look innovative is causing massive operational inefficiencies and leaving employees entirely confused.

We have watched tech companies push half-baked software for years, but the current generative AI gold rush is reaching peak corporate absurdity.

The Costly Rush to Look Innovative

Take the case of an AI engineer named Malcolm, who spoke to the BBC about his time at a data analysis firm. His bosses wanted to use generative AI to categorize a customer database into various personas.

Malcolm advised against it, pointing out that a traditional machine learning model would be cheaper, more accurate, and produce repeatable results. Naturally, the executives ignored the expert and went with generative AI anyway just to say they did.

The result was a slower, more expensive, and less accurate process. But hey, at least the C-suite got to brag about embracing AI in their next slide deck.

Mandatory Metrics and Tracking Dashboards

Other firms are taking an even more heavy-handed approach by forcing adoption through performance metrics. In February, Accenture reportedly told its staff that promotions to top roles would require regular use of its in-house AI tools, which the company is actively tracking.

KPMG went a step further in May, launching a dashboard to track whether US employees hit a 75 percent usage target for its AI tools. KPMG claims this is part of a holistic effort to help people move up the AI maturity curve, which sounds like classic corporate speak for “use it or else.”

While organizations like Anthropic are making headlines as Anthropic Files to Go Public, Setting Stage for Huge I.P.O., the actual implementation of these tools on the ground remains a mess.

Disconnect at the Top and Broken Cultures

It is not just the private sector failing to coordinate this transition. The UK government is also banking on AI to rewire the state and boost efficiency across Whitehall.

However, a survey by the civil servant union, the FDA, revealed that less than a third of civil servants were actually consulted on how the technology would be rolled out. FDA General Secretary Dave Penman noted that this inconsistent rollout across departments is actively limiting any potential productivity gains.

Dan Boyles, the CEO of consultancy Hello AI Collective, shared a story that perfectly sums up this executive blindness. He sat down with the C-suite of an oil and gas company and asked why they wanted to use AI.

None of them could agree on the goal. The CEO wanted to keep up with competitors, the head of sales wanted more money, and marketing wanted to fire their outside contractors.

Eventually, the president admitted he just wanted to boost operating earnings so he could sell the company in a few years. Once Boyles had that actual motivation, his team could finally locate real bottlenecks and apply AI where it actually made sense.

A bad company culture will quickly sink any AI rollout, according to Caroline Rawlinson, CEO of Culture Amp. Her firm found that while nine out of ten HR professionals expect to increase generative AI use, a third admitted that nobody actually owns the AI strategy at their company.

Rawlinson warned that putting AI on top of a fragmented or fear-based culture is a recipe for a slow, wasted effort. I agree entirely; you cannot fix a broken corporate culture by throwing expensive algorithms at it.

The Path to Real AI Integration

To make these systems work, companies need to focus on transparency and explainability. An eBook by Pega highlights that building trust in AI-powered decisioning requires breaking down data silos and creating explainable AI models.

An unnamed senior consultant at a major firm told the BBC that their employees must undergo mandatory training on AI ethics, bias, and hallucinations before getting tool access. Some of their staff use up to five different AI tools, but only after they understand that these systems can be sycophantic and outright