Beyond Skills: What AI Needs to Understand About People

For decades, we have watched our profession continually expand its understanding of people at work.

We began by standardizing competencies to create more consistent expectations for performance. We then moved toward skills, providing more granular descriptions of what people know and can do. As technology matured, those skills became interconnected through Skills Intelligence, making it easier to match people to jobs, projects, learning opportunities, and career paths. More recently, Talent Intelligence has combined multiple sources of workforce data to improve hiring, development, succession planning, and other talent decisions.

Each advancement has been meaningful. None fully replaced the generation before it. Each expanded our understanding and/or our ability to apply it.

Today, AI presents another opportunity—not to replace what we’ve learned, but to build upon it.

The Evolution of Our Understanding

This graphic isn’t intended to capture every innovation in HR technology. Workforce intelligence, organizational intelligence, people analytics, and many other disciplines continue to evolve alongside this progression.

Rather, it illustrates one important dimension of our profession: the expanding understanding of individual human capability.

Competencies standardized expectations.
Skills described capability. The What.
Skills Intelligence connected capability. The Map.
Talent Intelligence improved decisions. The Insight.
Human Capability Intelligence helps explain why people succeed and provides evidence-based guidance for growth. The Why.

Why does this matter to Talent Management?

Because AI is increasingly being asked to coach employees, guide development, support managers, identify potential, recommend career opportunities, improve teams, and inform succession decisions.

That raises an increasingly important question:

What does the AI actually know about people?

Skills Are Important—But What Do We Mean by “Skills”?

The conversation today is understandably centered on AI and skills.

Skills Intelligence has become foundational to workforce planning, talent acquisition, internal mobility, and learning. Organizations should absolutely continue investing in it.

But there is something I’ve been thinking about.

The word “skills” no longer means the same thing to everyone.

Depending on the platform, a skill may describe a technical capability, knowledge, a competency, a behavior, an attribute, a personality characteristic, or even a task. As our industry has expanded the definition, skills have become increasingly useful, but also increasingly ambiguous.

That isn’t necessarily a problem.

To me, it simply makes another question increasingly important: What actually enables people to succeed?

People Are More Than Their Skills

People do not arrive at work as collections of skills.

They bring knowledge, skills, behavioral attributes, motivations, experiences, aspirations, personality, judgment, learning agility, values, and relationships.

And none of these operates in isolation.

People work within teams, under different leaders, inside different cultures, and under changing business conditions. Who we are interacts with what we know, what we can do, the people around us, and the environment in which we’re being asked to perform.

Full human capability is complex.

What excites me about AI is that it may finally give us the ability to make that complexity more accessible, understandable, and useful at scale.

Is “Good Enough” Good Enough?

A few months ago, I was speaking with an organization planning to deploy enterprise AI as a talent resource for employees.

As we discussed the importance of grounding AI in validated talent science, someone made an observation that has stayed with me:

“The AI is good enough. Employees will have to discern and decide for themselves what’s right in the answers they receive.”

I understood the practicality behind the comment.

But I couldn’t help wondering:

Is “good enough” the standard we want for leadership guidance, coaching, career development, and succession decisions?

Our profession has spent decades building more valid, evidence-based approaches to understanding people and improving talent decisions.

If AI is going to influence those same decisions, shouldn’t we expect it to be grounded in that accumulated knowledge rather than asking employees to determine for themselves what guidance they should trust?

I believe our people deserve better.

We Started With the Science

This brings me to something I counted recently that I had never counted before.

Within TalentTelligent’s Human Capability Intelligence Layer are now more than 12,000 evidence-based relationships across human capability.

The number is interesting. But how we arrived there is more important.

We didn’t start with the data and ask AI to infer the relationships. We started with the science and worked backward.

We began with decades of validated, peer-reviewed talent management research and practical application around the knowledge, skills, attributes, and behaviors that contribute to effectiveness for leaders, managers, supervisors, individual contributors, high-potential talent, and teams. We combined that with decades of real-world experience, panels of leading experts, and ongoing statistical review from hundreds of thousands of 360 survey raters.

We then connected that knowledge across areas such as behavior, emotional intelligence, personality, coaching, development, developmental assignments, developmental difficulty, normative comparisons, potential, team effectiveness, and other dimensions of full human capability at work.

Yes, it’s complex.

Hot take: Humans are complex.

But AI gives us an opportunity to make that complexity accessible and useful in ways that simply weren’t practical before.

The 12,000+ relationships aren’t the important story by themselves. They represent decades of accumulated Talent Management knowledge that AI can now reason across.

That knowledge didn’t emerge from enterprise documents. It didn’t come from scraping the internet. And it didn’t simply present itself through fine-tuning. It came from decades of research, validation, and practical application.

LLMs + Enterprise Data + Human Capability Intelligence

Large language models have fundamentally changed how we interact with information.

Organizations are now enriching those models with their own documents, policies, processes, and operational data, giving AI valuable organizational context.

Those are remarkable advances. But I believe there is another layer that deserves equal attention.

An LLM understands language. Enterprise data provides organizational context. Neither, by itself, teaches AI the science of people. That requires a knowledge layer grounded in validated relationships about human capability.

Same Skills. Different Outcomes. Why?

Consider two people. Same job. Similar education, experience, and skills. Very different outcomes.

Why?

Anyone who has worked in Talent Management has seen this play out. Skills tell us a great deal about what someone knows and can do, but they don’t necessarily explain the whole person, or the context in which that person is being asked to perform.

Now consider what that could mean for AI.

Imagine a newly promoted manager asks their organization’s AI assistant:

“One of my strongest employees is missing deadlines. I don’t want to micromanage them, but I need to address it. How should I handle the conversation?”

A general-purpose AI can generate a reasonable response: clarify expectations, ask questions, listen, agree on next steps, and follow up.

There’s nothing inherently wrong with that advice. But what if the AI could go deeper?

What if it could draw on validated, grounded relationships among managerial behaviors, individual attributes, performance, development, and situational factors?

Instead of immediately prescribing an answer, it might help the manager diagnose the situation:

Is this an expectations problem? A capability gap? A delegation issue? A workload problem? Has decision authority been clearly established? Is the manager avoiding accountability because of discomfort with conflict? Does the employee need development rather than correction?

And, where appropriate and permissioned, what if the AI could also consider relevant organizational context; role expectations, development goals, assessment insights, or 360 feedback?

Now the interaction begins to change from:

“Here is some generally accepted advice about managing performance.”

to:

“Given this person, this manager, this situation, and what validated Talent Management research tells us, here are the factors worth considering and why.”

That’s the opportunity I see in Human Capability Intelligence.

AI shouldn’t replace managerial judgment or make consequential talent decisions for us. It can make better evidence available at the moment judgment is required.

An LLM enables the conversation. Enterprise data provides organizational context. Grounded Human Capability Intelligence brings an evidence-based understanding of people.

Together, they create the potential for AI to do more than provide answers that sound reasonable. They can help put decades of Talent Management science into the hands of managers and employees at the moments they need it.

This is one reason we made our Intelligence Layer accessible via API: to enable organizations to connect evidence-based Human Capability knowledge directly to the AI systems and workflows they are already building.

The goal isn’t simply an answer that sounds right. It’s guidance grounded in what we actually know about people, performance, and development.

Looking Ahead

Every generation of HR technology has expanded what organizations can understand about people.

AI gives us an unprecedented opportunity to operationalize decades of evidence-based Talent Management knowledge in ways that were never before possible.

The opportunity isn’t to abandon skills, competencies, assessments, or everything our profession has learned. It is to connect that knowledge and make it more accessible, explainable, and useful at the moments when people and organizations need it.

Perhaps the future of HR AI won’t be defined by who has the largest model or the most data.

Perhaps it will be defined by who has the deepest, most validated understanding of people.