Employers want great soft skills. Can they actually measure them?
5 minute read
5 minute read
Andreea Zaman
•
Andreea Zaman
•
The WEF Future of Jobs Report tells us what skills employers need. Tech leaders are saying human skills will matter more in the age of AI, not less. Both claims may be right. Neither tells us how to measure, teach, or reward those skills in practice.
The WEF Future of Jobs Report tells us what skills employers need. Tech leaders are saying human skills will matter more in the age of AI, not less. Both claims may be right. Neither tells us how to measure, teach, or reward those skills in practice.
The World Economic Forum's Future of Jobs Report 2025 surveyed over 1,000 employers across 55 economies, representing 14 million workers. The top skills they identified are analytical thinking, resilience and flexibility, leadership, creative thinking, motivation, and self-awareness.
These are not technical skills. They are cognitive and behavioural capacities that are genuinely difficult to measure well.
This view is not confined to the report. Daniela Amodei has argued that the qualities that make us human, including communication, curiosity, emotional intelligence, and compassion, will matter more at work in the age of AI, not less. This echoes what Ginni Rometty has said about generative AI placing a premium on collaboration and critical thinking, and what Satya Nadella has pointed to when discussing empathy and relational skills as AI absorbs more technical work.
There is something worth taking seriously in this consensus. But I think it skips a harder question.
These skills have always mattered. The problem is that we have rarely treated them as core hiring criteria in a way that is transparent or defensible. We signal them. We list them in job descriptions. We score them in interviews using methods that were never validated for that purpose. We rarely ask whether the assessment captures the construct, whether scores are reliable, or whether they predict anything meaningful on the job.
This is the measurement gap. Psychological science has spent decades building rigorous frameworks for assessing exactly these capacities. Many of the tools organisations use to hire and develop people were not built on those foundations, and in most cases were never validated against them.
The question that follows from the Amodei, Rometty, and Nadella framing is not whether human skills matter. It is whether we know how to measure them, teach them, and reward them. Right now, for the most part, we do not have convincing answers to any of those three.
The stakes are rising. Entry-level opportunities are tightening, at least in the UK data I have been tracking, which changes the risk calculus for young people choosing a degree or a career path. If human skills are genuinely what will be most valued, then those entering the workforce deserve to know that the systems designed to assess and develop those skills are actually fit for purpose.
The EU AI Act will force some of this into the open, at least for hiring tools classified as high-risk under Annex III. The broader question, across hiring, edtech, and performance management, is not whether better measurement is possible. It is. The question is whether organisations will wait for regulation to require it, or start asking now what their assessments are actually measuring.
The World Economic Forum's Future of Jobs Report 2025 surveyed over 1,000 employers across 55 economies, representing 14 million workers. The top skills they identified are analytical thinking, resilience and flexibility, leadership, creative thinking, motivation, and self-awareness.
These are not technical skills. They are cognitive and behavioural capacities that are genuinely difficult to measure well.
This view is not confined to the report. Daniela Amodei has argued that the qualities that make us human, including communication, curiosity, emotional intelligence, and compassion, will matter more at work in the age of AI, not less. This echoes what Ginni Rometty has said about generative AI placing a premium on collaboration and critical thinking, and what Satya Nadella has pointed to when discussing empathy and relational skills as AI absorbs more technical work.
There is something worth taking seriously in this consensus. But I think it skips a harder question.
These skills have always mattered. The problem is that we have rarely treated them as core hiring criteria in a way that is transparent or defensible. We signal them. We list them in job descriptions. We score them in interviews using methods that were never validated for that purpose. We rarely ask whether the assessment captures the construct, whether scores are reliable, or whether they predict anything meaningful on the job.
This is the measurement gap. Psychological science has spent decades building rigorous frameworks for assessing exactly these capacities. Many of the tools organisations use to hire and develop people were not built on those foundations, and in most cases were never validated against them.
The question that follows from the Amodei, Rometty, and Nadella framing is not whether human skills matter. It is whether we know how to measure them, teach them, and reward them. Right now, for the most part, we do not have convincing answers to any of those three.
The stakes are rising. Entry-level opportunities are tightening, at least in the UK data I have been tracking, which changes the risk calculus for young people choosing a degree or a career path. If human skills are genuinely what will be most valued, then those entering the workforce deserve to know that the systems designed to assess and develop those skills are actually fit for purpose.
The EU AI Act will force some of this into the open, at least for hiring tools classified as high-risk under Annex III. The broader question, across hiring, edtech, and performance management, is not whether better measurement is possible. It is. The question is whether organisations will wait for regulation to require it, or start asking now what their assessments are actually measuring.