AI isn’t broken — leadership is wasting its potential

For years, the big argument was whether AI actually worked. That question is largely settled. Budgets have been allocated, tools rolled out, and pilots pushed into production.

And yet the payoff is underwhelming.

The ROI gap leaders don’t want to see

According to IBM’s Institute for Business Value, only 25% of AI initiatives have delivered their expected ROI. Just 16% have successfully scaled across the business, despite years of investment and genuine enthusiasm.

Recommended reading

Mira Murati’s lab unveils Inkling, a 975B open model

The piece argues the core issue is not the technology, but the organizational structures around it. That makes it a leadership problem, not a tooling problem.

Spending on the wrong things

The default playbook has been predictable: buy platforms, spin up pilots, bring in a vendor for one-off training. That may solve tactical issues, but it sidesteps the main constraint — the “human infrastructure” required to make AI useful.

The companies seeing the strongest outcomes are not necessarily those with the most advanced or expensive models. They’re the ones that reengineer how their people work.

Boston Consulting Group’s analysis of AI leaders is telling:

• ~70% of resources go to people and process changes
• 20% to IT infrastructure
• 10% to the AI models themselves

Most organizations are running that ratio in reverse. When leaders obsess over tools and use cases, they underbuild the training, guardrails, and policies that make those tools reliable and repeatable.

Real productivity gains — with a catch

AI is already creating value for organizations that deploy it well. But those gains are far more fragile than many executives assume.

Usage is racing ahead of enablement. Among employees using AI at work, less than 8% say they’ve received extensive training on their tools — a number that has “barely budged” even as daily usage has climbed. 60% say it often takes longer to figure out how to complete a task with an AI tool than to just do it manually.

Companies are rolling out AI faster than people can learn to use it, risking exactly the kind of friction and confusion they were trying to remove.

What leaders actually owe their teams

Closing the gap between AI’s potential and its reality won’t happen through new procurement cycles or splashy rollout announcements. The author argues it demands ongoing, deliberate investment in people, centered on three shifts:

1. Train people, not just tools

AI capabilities are changing faster than static training decks. Instead of only teaching features, leaders should invest in role-specific judgment: where AI accelerates work, where it introduces risk, and when human critical thinking has to dominate.

2. Stop worshiping adoption metrics

If 80% of the organization is “using AI” and productivity is flat or declining, adoption is the wrong KPI. Leaders should track time-to-completion on real tasks and accept what the data shows. Some use cases that look innovative on paper may simply be slowing people down and should be dropped.

3. Make AI governance operational, not checkbox

Treating AI policy as a compliance document isn’t enough. The companies getting this right bake governance into daily planning and execution, rather than an “acceptable use” PDF no one reads.

That includes leaders modeling where they use AI and where they don’t, and explaining why those boundaries exist.

Turning potential into reality

The tech is ready. What’s missing is leadership willing to decide when AI should and should not be used, what processes to rebuild instead of blindly automate, and how to support teams through those changes.

Until executives start asking those questions — and funding answers that center on people and process, not just models and platforms — the AI ROI gap is likely to remain exactly where it is.

The article appears as part of TechRadar Pro Perspectives, with the author’s views not necessarily representing those of TechRadar Pro or Future plc. TechRadar Pro also notes it has ranked the best HR software, and invites contributions via its Perspectives submission page.