IBM CHRO Nickle LaMoureax doesn’t consider herself an AI optimist. Instead, she calls herself an AI realist.
“AI is not magic. It’s amazing, impressive technology that can totally transform your business,” she told a packed room of attendees during her opening keynote at Tuesday’s HR Tech conference in Las Vegas. “But it takes hard work, behavior change, culture change, business process change and sometimes leadership change.”
LaMoreaux, HR Executive’s 2024 HR Executive of the Year, experienced all of those realities as she led IBM’s AI in HR journey over the last decade—one that has been rife with both progress and pitfalls.
“It was a rocky road. It was anything but amazing,” says LaMoreaux, who has been with IBM for 25 years and now leads all HR for all 300,000-plus employees across 170 countries.
The tech giant started down the AI in HR path in 2017, when it gave all employees access to Watson Assistant. Before they knew it, employees had turned into “citizen developers,” and chatbots were “running around” everywhere. HR alone had 30 different chatbots, with very limited governance.
That prompted LaMoreaux’s team to centralize all of those tools into one solution: AskHR, which they envisioned as a one-stop shop for all things HR. But there was one big problem: No one was using it.
So, the team made the “bold” decision to force the change: Overnight, they shut off the HR email address and phone number and told 21,000 frontline managers they no longer had access to their HR business partners. All of their HR questions would have to go through AskHR.
RELATED: More from HR Tech 2025
Unsurprisingly, LaMoreaux said, the move didn’t go over too well with IBMers. Before that decision, HR’s employee Net Promoter Score sat at +19; after, it dropped to -35.
But the team persisted, she said. With budgeting challenges, an increasingly regulated environment and evolving employee expectations for tech experiences at work, “we knew AI was our only way out of this,” she said.
Listening and learning
LaMoreaux’s team consistently learned from AskHR—what was and wasn’t working, including data culled from interactions that got a “thumbs down”—kept leadership informed and focused on continuous improvement. At the same time, the AI was getting smarter and HR shifted its operating model: Funneling complex AskHR requests to skilled experts—not in traditional HR domains like talent acquisition but in areas that could bring more meaningful support to employees and managers. For instance, a manager consulting AskHR about a low-performing employee may be passed along to an HR expert specifically skilled in creating learning plans.
“We created these cross-functional experts to answer questions from the point of view not of HR silos but of what the problem was,” she said.
HR also revamped the solution to embrace agentic AI—treating AskHR as an orchestration layer for all of IBM’s HR systems—enabling more seamless transactions.
Today, AskHR is trained in 52 languages and 30 HR domains. It handles 11.5 million transactions every year—with 100% adoption by managers and 99% adoption by executives. Currently, 94% of employee questions are handled entirely in the system. Last year, half of employees and managers never logged into an HR system at all—and LaMoreaux said she’s on the path to getting that figure to 100%. The dramatic uptake in adoption of AskHR has led to a 40% reduction in the HR operating budget.
And today, HR’s eNPS sits at +74.
See also: What will it take to be a true ‘future-ready’ org?
5 lessons from IBM’s AI in HR journey
Such a turnaround didn’t come quickly—or easily. But it did convey a number of meaningful insights about what it takes to build an AI-first HR function successfully. Among them:
Forget FOMO
Don’t follow “best practices” for integrating agentic or generative AI—because they don’t exist, LaMoreaux said. Instead, focus on whether the AI supports your particular business model, talent model and culture. AI shouldn’t be deployed everywhere. For instance, IBM is “pretty far advanced” in using it for compensation practices, but still in the early stages when it comes to talent acquisition. “Just because another company or your competitor is doing it,” she said, “doesn’t make it right for you.”
Start small and targeted
High-volume transactions are usually a good place to start integrating AI in HR. Drive even more meaning by focusing on “moments that matter.” For instance, at IBM, AI was able to cut the delivery of employment verification letters from up to two days to about 30 seconds, which made a significant difference for employees. LaMoreaux also advised HR to focus on the process or policy that draws the most complaints. “You’re not going to make it worse,” she laughed. For IBM, that was employee transfers, a “dreaded” and complex process performed in HRIS; AskHR now handles all 80,000 annual transfers, with a near error-free record.
Fix processes first
Throwing AI on top of bad processes won’t improve them—but it will waste time and money. LaMoreaux followed a simple framework for process improvement: eliminate, simplify, automate. “Before anything went into AI, you had to prove why it needed to exist at all,” she said.
Be user-centric
As HR builds AI into the function, rethink overall operating models to ensure employee and manager needs are at the heart, LaMoreaux said. For instance, are there mechanisms to ensure feedback about the tech?
Prioritize culture
HR functions can deploy the best technology in the world—but without the culture to support it, the effort will fall flat, she said. For instance, IBM had the culture that enabled her HR team to ride out that -35 eNPS, but not all organizations will—so, understand the culture in which you’re operating before advancing on your AI in HR journey.
HR’s responsibility as an AI ‘myth-buster’
As HR leads this work, it may need to contend with disruptive perspectives. It will be up to HR to bust “myths” like:
1. Domain expertise is no longer critical.
Both for the HR function and companywide, HR should actually “double down” on the value of domain expertise. AI will do the basics, but it will never have the judgment, wisdom and experience human employees bring. “I tell my HR employees, ‘You need to be one step smarter than AI at all times,’ ” LaMoreaux said.
2. The early career ladder is broken.
Dialing back entry-level hiring as AI takes over tasks may make a CFO happy for a year—but what about a few years down the line, when mid-level experience is needed?
“Where is that supposed to come from?” LaMoreaux questioned. She actually needs more entry-level HR professionals than before IBM deployed AskHR—they’re just doing different work: training to answer complex questions or interacting directly with employees who report a negative experience with the tech.
3. AI runs wild.
It is “critical” to have a mechanism to monitor your AI—so you know in real time what’s happening and can catch hallucinations or problems quickly. “You are in the driver’s seat about when, where and how you use AI,” she said. “Whether or not AI runs wild is really in your hands.”
4. Massive job loss is inevitable.
Headlines abound on this topic, LaMoreaux said—but it doesn’t “compute” to her. AI will only take jobs at organizations where growth isn’t expected by investors, shareholders and other stakeholders. “But the pie is growing” at most organizations, she said. “AI is doing some work, but it’s important for us to think about redirecting employees to where the growth is in your organization. [AI-driven job loss] is a major fear—whether people admit it or not—in most organizations that is going to get in the way of AI adoption.”
A new moment for HR
LaMoreaux urged attendees to be bold when it comes to AI in HR: Whatever the plan is for integrating the tech, push it a step further.
And, think outside of the function. HR has the opportunity to not only deploy AI strategically in their own work but across the enterprise.
“It’s your time to be in the sun,” she said.