Employers Rethink Skills Validation as AI Reshapes Jobs

Employers are rethinking how they identify, assess, and validate talent as AI reshapes jobs and accelerates changes in workplace skill requirements.

Kasey Guentert, an industrial-organizational psychologist and head of skills advisory at Vervoe, sat down with SHRM at the HR Tech 2026 conference in Las Vegas to discuss how AI can uncover overlooked capabilities, help organizations identify workforce development needs, and benchmark skills against industry standards. She also explained why ongoing assessments, human judgment, and actionable employee feedback remain essential.

Specialty Credential: AI+HI

SHRM: How can employers use AI to identify skills that traditional hiring processes might overlook?

Guentert: AI gives us access to data in a way that never existed before. That’s both an opportunity and a risk, because you don’t always know where the data is coming from or how reliable it is. That’s why benchmarking is key. With well-designed prompts, you can direct a large language model (LLM) to the most relevant sources, such as job postings for comparable roles across your industry, and use them to build a clear picture of which skills those jobs actually require. That tells you what to measure, so you can measure it well.

For example, a salesperson at one large CRM in San Francisco and a salesperson at a different large CRM in the same city should need broadly similar skills. You want to make sure the skills you’re hiring for match what matters most in the industry, not just what’s in your own job description. AI makes that possible because so much job information is publicly available, and it can comb through far more of it than any person could. The output still needs expert review, but it’s a much stronger starting point.

SHRM: How can employers use skills data to uncover hidden capabilities within their existing workforce?

Guentert: This comes up a lot with our clients. Many want to uncover hidden skills, and ideally that’s what assessment does: it shows where people can grow and develop, rather than judging them only on past experience. Resumes and job histories tell you what someone has done. They don’t tell you what someone can do next.

Most people aren’t moving into entirely new jobs. Their current jobs are changing in the age of AI. So the question becomes: How do we make sure people don’t fall behind? Skills testing is one of the best ways to answer it. Employers start with an inventory of what future roles will require, then test employees against it to verify who already has those capabilities. Just as importantly, it shows who doesn’t have them yet, so the organization can invest in development before it becomes a performance problem. And because it’s done at scale, we can summarize the results and show organizations the trends across teams and roles.

SHRM: What has the employer response been for these talent assessment use cases?

Guentert: We’ve seen a range. At the best end, there’s real planning involved. Some clients have identified all the skills each job needs, verified that list with the employees doing the work, and had employees self-assess. Then they’ve used the results to help people move into new roles or build the skills they need through learning and development. That’s amazing, and I’d like to see every organization do it.

At the other end, some companies assess employees against a skills library but never share feedback with them, often because they aren’t sure what to do with the data. Some talent teams simply don’t have the internal capacity to act on the results, and that’s where I’d like to see more partners step in to help. Whatever you do, don’t leave employees wondering what an assessment means for their performance or their development.

Research: The State of AI in HR 2026

SHRM: As AI changes jobs faster than traditional competency models can keep up, how should employers rethink skills validation? How frequently should skills be reassessed?

Guentert: In 2015, I would have said every two years. My current recommendation is to reassess about every six months, or even more often if the job is changing quickly. Ideally, the cadence is driven by data: benchmarking each role over time tells you how fast its requirements are shifting. Our philosophy is to look at how people actually do the things the job requires. AI fluency, for example, has become a big request from our clients. They want to see how well people can work alongside AI agents and LLMs. The goal is to catch early indicators of strengths and development opportunities before gaps show up in performance.

SHRM: Is the old way of creating and managing skills competencies extinct?

Guentert: For jobs undergoing frequent change, yes, I think that approach is over. But that isn’t true for every job. For roles whose core requirements change slowly, such as police officers, librarians, teachers, and construction jobs, traditional competency models are still useful and reasonable.

SHRM: How should employers balance AI-generated skills recommendations with human judgment?

Guentert: Generally, use AI to see what comes up: ask it the right questions, then evaluate its responses very carefully. There are areas where AI can handle a task from start to finish on its own, but in HR that’s rare, because these decisions affect people’s livelihoods.

At Vervoe, we don’t use generative AI or LLMs to score candidates. Our scoring uses machine learning models, because right now LLMs can’t score consistently enough for hiring decisions. Even if a scoring mechanism claims 85% correlation with human scoring, our concern is that the remaining 15% is massively off or causing a risk.

SHRM: In the next few years, what will leading employers be doing to validate skills that most organizations aren’t doing today?

Guentert: The big trend is transportability: applying validation evidence gathered elsewhere to a new setting when the jobs are similar enough. For example, if a job analysis shows a role is a call center representative handling collections, we can draw on assessments already validated for that same type of role. That evidence has held up time and again, so rather than starting from scratch, we map a similar situation onto proven evidence. AI will make it faster and easier to establish that two jobs are truly comparable, which is the foundation transportability depends on.