The people function must lead AI transformation: Here are five fresh ways to do it

Here’s an uncomfortable truth: 66% of HR leaders say their companies are ‘operational’ with AI, yet rate their own confidence in leading company-wide AI strategy at just three out of five.

In interviews and surveys I conducted with 50+ HR leaders at some of the world’s largest employers, they cite many obstacles – unclear C-suite vision, uncertainty around ROI, and ‘difficulty connecting AI initiatives across the organization’.

The pattern between these is telling. Getting value out of AI isn’t a problem with the technology itself. It’s a people problem. That means HR should be in a position to solve it.

Experts agree. Ethan Mollick, Wharton professor and bestselling author of “Co-Intelligence,” describes HR as ‘R&D’ when it comes to AI adoption.

The capabilities that make great HR professionals – curiosity, conversation, and understanding how humans and organizations learn and react to incentives – are exactly what can unlock AI’s practical potential.

And yet, only 30% of the HR leaders I spoke to believe their CEO would say HR is the right function to lead AI transformation.

I believe part of this low confidence stems from the HR function not having an informed toolkit of approaches they can adapt and apply.

Encouragingly, some HR leaders I interviewed are taking unique and promising initiatives to prepare for AI’s impact.

Here are five that stood out.

1. Making AI skills universal

Some HR leaders are preparing for an AI-driven workforce by tackling the skills gap at the source, i.e. when new people join.

At Allianz, Global Head of Learning & Skills Management Isabelle Kokoschka has made Gen AI a mandatory skill.

In part, this means that all job descriptions now carry a version of this requirement, and the company has challenged assumptions about which roles need AI skills, including, for example, call center staff.

Allianz has also gone further, embedding AI into daily habits with practical learning nuggets like ‘Reduce Your Ugly Four Hours’ to identify and eliminate inefficient tasks that AI can handle.

Some leaders ask ‘did you use AI?’ at the end of every meeting.

These simple rituals normalize adoption far more effectively than formal training programs.

2. Becoming a ‘teaching organization’

For new employees joining the workforce for the first time, AI presents a particularly difficult conundrum.

Many of the basic tasks that helped new employees learn critical skills like writing, reasoning, and collaboration are being automated by AI.

What’s more, remote work – still prevalent at many companies – makes it harder to learn through osmosis.

The effect of this double challenge is real. Margaret Burke, Talent Acquisition & Development Leader at PwC US, noted that the firm could see “promotion timelines starting to elongate”.

PwC US has responded by launching a firmwide ‘apprenticeship model’ across 75,000+ employees as part of a drive to become a ‘teaching organization’.

The program is built around five regular rituals:

• model and shadow behaviors
• ask and answer questions
• share and seek knowledge
• give and request feedback
• stretch and grow skills

These are embedded throughout employee evaluation forms, formal feedback programs, surveys, and recognition programs.

3. Deliberately make space for expertise

Recent research from David Autor and Neil Thompson reveals an important dynamic about how the impact of AI may play out on the labor market.

Their paper shows that when the more expert part of a job is automated, the remaining job becomes accessible to more people and wages fall.

Conversely, when the more ‘boring’ parts of a job are automated and the expert part remains, fewer people are able to take on that job and its value increases.

Mikala Larsen, Head of Corporate Learning, Development & Leadership at Nestle, intuitively understands this dynamic.

The senior leadership programs that her team builds actively embrace AI on one hand – for  example showing how some core work can be compressed from 4 months into 2.5 days.

But on the other hand, the same programs deliberately create “space for ethical reflection” and “maintaining space for human expert decision-making despite AI’s speed”.

4. Build internal talent marketplaces

It’s difficult to be sure of when and how AI will affect jobs. One way to prepare is for organizations to make it easier for employees to move to jobs that are in greater demand.

At the Swedish manufacturer Atlas Copco Group, CHRO Cecilia Sandberg is launching a ‘skills-based talent transformation’.

Her bet is that by grounding talent acquisition, internal mobility, and learning and development in more objectively defined skills and capabilities the group will make more effective use of its globally distributed workforce.

Only a few months in, the organization is scoring early wins including more diverse candidates and reduced ‘time to hire’.

One element that’s been crucial to this setup is the people function’s strong relationship with stakeholders across other functions and business units.

According to Sandberg, “90% of HR work involves stakeholder collaboration”.

This is an important insight for all HR leaders, particularly as my research shows that only 30% of HR leaders believe their CEO trusts them to lead AI transformation.

5. Create new organizational structures

Some HR teams are embracing new organizational structures to prepare for an AI-driven workforce.

Swarovski created a “cross-functional AI responsibility community” where the Chief Digital Officer and CHRO collaborate, tracking progress through AI scorecards.

Lloyds Banking Group created a ‘data & AI culture’ team within their data and digital function with responsibility for training and skills across the group .

More famously, the pharmaceutical company Moderna merged its HR and IT functions entirely.

These new approaches are getting results. For example, L’Oréal built a small AI task force with dual reporting into IT and HR, and credits this team in part with helping the company rapidly achieve over 50% adoption of Copilot.

These structures are symbiotic – HR needs IT expertise to select and deploy the right AI tools, and IT needs HR to ensure employees know where and how to use them.

Conclusion

The bottleneck in AI transformation isn’t models or licenses – it’s effective adoption.

This is precisely HR’s domain: changing behavior at scale, codifying new practices, aligning incentives, building organizational capability.

The irony is hard to miss: the function least trusted to lead AI transformation (by its own admission) may be the only one capable of making it work.