How HR Leaders Should Be Using AI in 2026

HR Is Using AI to Polish What Already Exists. That Is Not What AI Is For.

The function that is supposed to help organizations prepare for the future is spending most of its AI investment optimizing the past.

There is a specific pattern in how organizations adopt new technology, and it is almost always the same. The technology arrives. People figure out how to use it to do the things they were already doing faster or more cheaply. They declare success. And then, somewhere in the distance, someone who understood the technology differently has used it to do something that the first group had not considered possible, and the gap opens.

In the HR function, Dean Carter — former CHRO at Patagonia, Sears, and Fossil, now CEO of an AI company in the HR industry — argues that this gap is already forming, and that most HR leaders are on the wrong side of it.

“HR is navel-gazing,” he said recently. “We are applying AI to what already exists. Instead, we should be looking out at the totally new horizon that is being created by AI.”

The critique is specific and worth unpacking carefully, because it points to something real about how the function tends to approach its own transformation.

•   •   •

The Optimization Trap

The most common AI applications in HR right now are workflow automations. Job description generation. CV screening. Onboarding documentation. Interview scheduling. Performance review templates. These are legitimate uses of the technology. They reduce time spent on administrative tasks, they increase consistency, and they free up HR professionals for higher-value work, at least in theory.

Carter’s argument is not that these applications are wrong. It is that they represent the minimum viable use of a technology with significantly larger capabilities, and that HR functions that stop here are treating AI as a better typewriter rather than as a fundamentally different tool.

The analogy that comes to mind is useful: the first people to use the internet built electronic versions of newspapers. The people who built what the internet actually became did something different. They asked what was possible that had not been possible before, rather than how to do existing things more efficiently.

For HR, the existing-thing-more-efficiently version is writing better job descriptions in thirty seconds instead of an hour. The new-thing version is what Carter describes at his own company: a system that passively analyses observable behaviours in virtual meetings across an entire organization, identifies the patterns that correlate with team performance and individual thriving, and delivers personalized insights to every participant after every meeting. This would have been operationally impossible without AI. It is not a faster version of a survey. It is a category of organizational intelligence that did not exist.

•   •   •

The 80/20 Principle of AI Use

Carter’s framework for how to actually use AI is worth adopting directly. His version: AI delivers about 80 percent of what you need. That 80 percent handles the knowledge collection, the research legwork, the synthesis of information that takes more time than it takes brainpower. The remaining 20 percent — the judgment, the nuance, the decisions that require context and wisdom that no model currently holds — is where human expertise should be concentrated.

The mistake is expecting the model to deliver 100 percent, which it cannot, and then concluding either that AI is not ready or that your role is to prompt it into increasingly refined versions of a complete answer. Neither response is useful. The first undersells the technology. The second wastes the human.

For HR specifically, this means being clear about which parts of the function require genuine judgment — the interpretation of cultural signals, the handling of sensitive employee situations, the strategic alignment of people decisions with business direction — and using AI aggressively on everything that does not. The recalibration is uncomfortable for functions that have historically justified their value through process ownership. When the process is automated, the question of where HR’s value actually lies becomes more urgent and more visible.

•   •   •

How to Build AI Literacy in a Team

Carter’s approach to building AI competency within his teams offers a model directly applicable to HR leaders managing their own functions.

The starting point is play, not policy. His recommendation (that employers reimburse AI tool subscriptions for employees to use at home) is not primarily about cost. It is about the quality of learning that happens when someone has unstructured time to experiment with a tool versus structured time to apply it to a specific task. People who play with AI at home before using it professionally develop a more accurate intuition about its capabilities and limits than those who encounter it only in the context of a defined workflow.

The policy-first approach to AI adoption: governance frameworks, approved tools lists, use-case documentation before deployment, is not wrong, but it can become its own obstacle when the governance structure moves more slowly than the technology. Carter makes a specific distinction: if an external vendor uses AI to deliver a solution, and that AI does not touch the organization’s systems, it should not be subject to the same governance review as internal AI deployment. Organizations that apply the same level of oversight to both create a drag on adoption that the technology does not require.

•   •   •

The Transparency Problem

One of the more honest observations in Carter’s thinking concerns the gap between how AI adoption is framed by leadership and how it is experienced by employees. When leaders ask teams to break down their work into discrete tasks so the organization can assess where AI adds value, the employees hear one thing regardless of how the question is phrased: headcount reduction is coming, and this exercise is the map.

Carter’s response to this is not to avoid the topic but to be transparent about it. “Be transparent and have a strong reskilling program for those impacted.” This sounds obvious and is, in practice, rarer than it should be. Organizations that manage AI adoption through opaque restructuring and framing exercises tend to produce exactly the fear and resistance they were trying to avoid. Organizations that name the change, its scope, and the investment they are making in the people affected by it tend to get more engaged participation in the transition.

The principle is the same one that applies to most organizational change: people can handle difficult truths significantly better than they can handle the suspicion that the truth is being managed.

•   •   •

What This Means for African HR Leaders

The conversation about AI in HR tends to occur in a context centered on organizations in the US and UK, with enterprise infrastructure and AI tool budgets that do not map directly onto the Nigerian or broader African context. The $20-a-month AI subscription that a US CHRO recommends reimbursing is not a trivial line item for every HR function in Lagos.

But the underlying principle holds without adjustment: the organizations that will have a meaningful advantage in the next five years are those whose HR functions are building organizational intelligence rather than automating administrative tasks. The specific tools may differ. The orientation, toward what AI makes newly possible rather than toward what it makes faster, is universal.

The talent management, culture-building, and workforce planning challenges facing African organizations are no smaller than the ones facing Western organizations. In many cases, with higher attrition rates, more volatile economic conditions, and a workforce that is younger and more digitally native than almost anywhere else, they are larger. The HR function that learns to use AI to understand and anticipate those challenges, rather than to write better job descriptions faster, will be operating at a different level from the one that does not.

Carter’s closing provocation is worth sitting with: “People are hallucinating more than ChatGPT if they don’t think it will impact their roles.” The HR leaders best positioned to navigate that impact are the ones who stop looking inward at existing workflows and start looking outward at what the technology makes genuinely possible for the first time.

The horizon is new. The function that maps it will matter more than the one that polishes the old one.

Related: How to Use Claude Cowork as a CEO or Founder in 2026: A Guide for African Business Leaders

Related: How to Use Claude Cowork for HR in 2026: A Guide for African People Operations Teams