AI Pays Off When People Lead: Control, productivity, and trust in the future of work
Artificial intelligence is advancing faster than most organisations can absorb it. The technology is capable; the returns are not arriving. The reason is rarely the model. It is the absence of the governance, workforce redesign, and employee trust required to turn deployment into measurable value — and that is a leadership problem, not a technical one.
The productivity mirage
The first discipline leaders need is honesty about what AI is actually delivering. Perceived gains cannot be taken at face value. In a widely cited 2025 randomised trial by METR, experienced developers believed AI tools had made them roughly 20 percent faster — yet measured task times showed them running about 19 percent slower. METR has since cautioned that the finding may not reflect the latest tools, but the enduring lesson holds: self-reported productivity is an unreliable guide to real productivity.
For a CHRO, the implication is uncomfortable but clarifying. Enthusiasm is not evidence. Adoption dashboards, licence counts, and satisfaction surveys measure sentiment, not outcomes. Organisations that mistake activity for impact will keep funding tools that feel transformative while the operating numbers refuse to move. The discipline that separates leaders from laggards is the willingness to measure work before and after, against outcomes that matter — cycle time, quality, cost, customer experience — rather than asking people how much faster they feel.
Trust is the asset — and it is fragile
The second discipline is treating employee trust as a strategic asset rather than a communications afterthought. The starting position is unusually favourable. In McKinsey’s 2025 Superagency in the Workplace research, 71 percent of employees said they trust their employer to act ethically as it develops AI — ahead of universities, large technology companies, and start-ups — and 73 percent trust their employer to do the right thing in general, against just 45 percent for government. Few institutions command that kind of confidence.
That confidence is also brittle. The same workforces remain worried about fairness, job security, and being excluded from decisions that reshape their roles. What leadership frames as reallocation, employees often experience as a threat. Trust of this magnitude erodes quickly under opaque, top-down deployment. The remedy is candour: be explicit about where AI is being used, what outcomes are expected, and what it means for people’s work — and make unmistakably clear that change is being done with employees, not to them. Transparency is not a courtesy here; it is the mechanism that protects the organisation’s scarcest advantage.
Redesign the work, and draw the line of control
The third discipline is structural. Productivity gains do not come from layering AI onto processes designed for a pre-AI world; they come from redesigning the work itself. Speaking at UNLEASH World 2025, Josh Bersin observed that only about 5 percent of companies believe they are seeing a positive return on AI where people are concerned. The organisations pulling ahead are those rebuilding jobs, roles, and workflows around the technology rather than bolting it onto structures never built for it.
HR is the natural owner of that redesign. A workflow audit — led by HR, not IT — should map where AI is already in use, expose the gap between deployment and actual outcomes, and rebuild the affected workflows with people at the centre. The same function should draw the line of control. Control is not restriction; it is clarity about what still demands human judgement and what does not. A simple decision-rights framework does most of the work: for every process touched by AI, name who is accountable for the outcome, define what a human must review before anything is actioned, and specify what recourse exists when the system is wrong.
This is where the real differentiator sits. It is not AI spend, but the quality of the partnership deploying it. Speed will come from the technology function; durability will come from HR. The organisations that win will treat AI governance as a genuine HR–IT partnership — pairing IT’s pace with HR’s understanding of people — rather than handing it to either function alone. Put people in the lead, measure what is real, and protect the trust you already hold, and AI stops being an expense in search of a return. That is when it finally pays off.
How EX3 can help
EX3 exists to put technology in the service of people — and that is exactly the discipline AI adoption demands. We work alongside HR and technology leaders to close the gap between deploying AI and realising its value, across three fronts that mirror the argument above.
- Measure what is real. We run workflow audits that map where AI is already in use and quantify the gap between adoption and actual outcomes — replacing sentiment with evidence leaders can act on.
- Redesign the work. From global payroll transformation to skills-based organisation design, we rebuild roles, processes, and workflows around the technology — with people at the centre, not bolted on afterwards.
- Govern with confidence. We help you build the decision-rights frameworks and HR–IT operating model that make AI durable — clarifying accountability, human review, and recourse, and protecting the employee trust you already hold.
Contact us
