Emotional intelligence and AI: the 2026 data

Demand for emotional intelligence is rising, measured capacity is falling, and training budgets are shrinking. A close look at the research behind the buzzword, with every source and its limits.

"Emotional intelligence is becoming the key skill of the AI era." I read some version of that sentence almost every week in 2026, in press releases, in trade articles, in my LinkedIn feed. The largest HR fair in the German-speaking world, Zukunft Personal Europe in Cologne, even ran under the motto "Team Human x AI" this September. I agree with the sentence, otherwise I wouldn't do my work the way I do. But something about it bothered me: it's rarely backed up. A lot of pieces on the topic don't cite a single study, and the numbers that do get cited travel from article to article and lose their context along the way.

Since 2019 I've worked as a trainer and coach with teams from more than 30 DAX companies and SMEs. If you want to get emotional intelligence training approved inside a company, you need more than a good feeling about it, especially in a year when every budget line has to be justified. So we took a close look at the 2026 data: the big reports, the current research, and also the numbers that cut against the topic. You'll find the source for every figure here. And where a figure tends to get stretched, you'll find what it actually measures.

1. Demand: what employers are looking for in 2026

The most-quoted number of recent months comes from the Cornerstone Skills Economy Report 2026, published in December 2025. Cornerstone analysed 28 terabytes of labour market data from more than 200 countries and over 50,000 skills from 2023 to 2025. The result: demand for emotional intelligence rose by 95%, demand for resilience and flexibility by 42%, and demand for leadership and social influence by 28%. In Germany, Haufe was among those who reported on it.

One detail usually gets lost in the coverage: the 95% refers to advanced technical roles. So the figure doesn't apply to the labour market as a whole. To me that makes it even more telling. In the same report, demand for AI implementation skills rose by 245%. Exactly where AI is being built and rolled out, employers are increasingly looking for people who can also handle emotions, their own and their team's. Cornerstone calls this the "Great Skills Merge": the line between technical and human roles is dissolving.

The World Economic Forum paints a similar picture in its Future of Jobs Report 2025. Among the ten core skills employers worldwide name for their workforce, leadership and social influence rank third (61% of employers), motivation and self-awareness fifth (52%), and empathy and active listening seventh (50%). At the same time, employers expect 39% of their workers' core skills to change by 2030. 85% plan to prioritise upskilling their workforce.

How fast work is changing in Germany right now shows in a very fresh number: according to Bitkom, in September 2026, for the first time, more than half of companies with at least 20 employees use artificial intelligence, 57%. A year earlier it was 36%, two years earlier 20%. Every one of these rollouts changes roles, processes and responsibilities. And every change of that kind sets something off in people: curiosity, uncertainty, sometimes real fear about their own place.

AI use in German companies

The Deloitte 2026 Global Human Capital Trends report is titled "Choosing the human advantage" and is based on a survey of more than 9,000 business and HR leaders in 89 countries. Deloitte doesn't explicitly name emotional intelligence in it, and I want to be clear about that, because some summaries say otherwise. But the report describes exactly the gap where emotional intelligence does its work: 85% of respondents say building their workforce's ability to adapt is critical, yet only 7% say they're leading on it. And only 6% say they're making progress in designing how humans and AI work together.

How humans and AI work together is, in the end, a question of how people deal with change. That's where our programme Emotional Intelligence in Action starts: with what happens in teams when work changes faster than people can keep up.

2. What AI can do, and what it doesn't take over

One study stayed with me more than most this year. Katja Schlegel, Nils Sommer and Marcello Mortillaro from the universities of Bern and Geneva had six large language models, including ChatGPT-4, Claude and Gemini, take five established emotional intelligence tests. The result, published in Communications Psychology: the models averaged 81% correct answers. Humans in the original validation studies of these tests averaged 56%. ChatGPT-4 was even able to create new test items that worked about as well as the originals.

One qualification belongs here: the human comparison figures come from older validation samples, so this wasn't a head-to-head contest under identical conditions. The direction is still clear. These tests ask, for example, which emotion is likely in a described situation, or which response would be most helpful. That's knowledge about emotions, and a machine can clearly retrieve that knowledge.

When I look at what's actually hard in my workshops, participants almost never lack that knowledge. Most of them can explain very well what active listening is, or why it's better not to answer criticism in the heat of the moment. What's hard is the moment itself: when your own project gets taken apart in a meeting, your pulse goes up, and the answer is already on your tongue before any knowledge gets a chance. Those first seconds after a trigger, which I wrote about in the 90-second rule, happen in the body. No AI can live them for you, and no AI can make a different choice for you in exactly that moment.

The WEF supports this distinction. In an analysis by Indeed documented in the Future of Jobs Report 2025, skills rooted in human interaction, including empathy and active listening, currently show no substitution potential by generative AI. An honest note belongs here too: for this analysis, GPT-4o rated its own abilities. That's an interesting signal, but not independent proof.

For practice, we draw one clear conclusion. An emotional intelligence training that mainly teaches knowledge now competes with a tool that can retrieve that knowledge better than most people. What remains is the practice: noticing what's happening in you and in the room, understanding where it comes from, and deciding consciously how to go on. Those are the three steps of our PIN Method (Perceive, Investigate, Navigate), and they can only be practised, never just read.

3. The gap: the "Emotional Recession"

While demand is rising, measured capacity is going down. That's the finding of a study published in November 2025 in Frontiers in Psychology by Joshua Freedman and colleagues. They analysed data from 28,000 adults in 166 countries, drawn from a pool of almost 277,000 people who took a standardised emotional intelligence assessment between 2019 and 2024. The authors call their result an "Emotional Recession".

All eight measured competencies declined significantly. The steepest drop was in the area the study calls "Drive": intrinsic motivation, optimism and pursuing meaningful goals, down 7 to 8%. The other competencies fell by 4 to 6%. People with above-average emotional intelligence, on the other hand, were about ten times more likely to report strong overall life outcomes, across wellbeing, relationships and effectiveness.

The "Emotional Recession": decline per competency, 2019 to 2024

The Drive finding matches what I see in rooms. Motivation and meaning are exactly the capacities that protect people from exhaustion. When they fade, it rarely shows up as drama. It shows up as cameras that stay off, meetings where people have stopped pushing back, and an "it's fine" that actually means something else.

Other large data sources point the same way. According to Gallup's State of the Global Workplace 2026, only 20% of employees worldwide were engaged at work in 2025. Gallup estimates the cost of low engagement at around 10 trillion US dollars for the global economy. The decline is especially clear among managers: their engagement has dropped 9 percentage points since 2022. And Mercer reports in its Global Talent Trends that up to 82% of employees feel at risk of burnout.

Demand and capacity are measured differently in these studies, so the numbers can't be set off against each other. But they point in opposite directions. Organisations need more of a capacity that, on average, is currently shrinking. In my view, that gap is the real task for HR and leadership over the next few years.

4. The headwind: smaller budgets, more proof required

The 2026 data also has an uncomfortable side. According to the McKinsey HR-Monitor 2026, reported by Handelsblatt in May, German companies cut their training budgets by 30% over the past two years, to 1,204 euros per employee per year. Around 5,500 employees and 1,300 HR leaders in ten countries were surveyed. In Germany and the Netherlands, a third of respondents didn't spend a single hour in training last year. McKinsey publishes the full results in the HR-Monitor 2026.

So the emotional intelligence trend is real, and the money for it is contested. At the same time, a large share of the remaining budgets flows into AI training, because that's where the gap is most obvious.

For organisations, this means training becomes more selective. Fewer programmes get bought, and each one has to be better justified. An inspiring one-off day after which nothing changes in everyday work will be harder to defend in 2026 than a programme whose effect can be shown. I think that's a healthy development. It pushes every provider, including us, to demonstrate impact instead of just promising it.

If you have to defend a budget internally, you can turn the argument around: precisely because every training hour is getting scarcer, it should go into a capacity that AI doesn't take over and that is, on average, declining. We've gathered the arguments for this on the EIA page under "Why now".

5. Can emotional intelligence actually be trained?

It's a fair question: if emotional intelligence were a fixed part of someone's personality, every training would be wasted money. The best answer comes from a meta-analysis, a study that pools the results of many individual studies. Victoria Mattingly and Kurt Kraiger analysed 58 studies on emotional intelligence training for the Human Resource Management Review. Their result: training works, and the effect is moderate, neither tiny nor spectacular.

What does moderate mean in practice? The most reliable studies compared two groups: people who took a training and similar people who didn't. After the training, the average person in the training group scored higher than 67 out of 100 people in the group without training. If the training had no effect at all, it would be 50 out of 100, because an average person always sits right in the middle of a comparable group.

Researchers call this measure "Cohen's d", and in these studies it came to 0.45. For comparison, 0.2 counts as a small effect, 0.5 as a medium one and 0.8 as a large one. Studies that only measured the same people before and after the training found an even larger effect (d = 0.61, or 73 out of 100). That figure is less reliable, though: without a comparison group, you can't tell whether the change came from the training or from something else, for example people simply getting used to the test.

What d = 0.45 means in practice

One more qualification: published studies showed larger effects than unpublished ones, a common pattern in research, so the true effect may be a little smaller. The core finding still holds. Emotional intelligence can change, and it can be developed deliberately.

Our own numbers are smaller and less rigorous, but they're real. Emotional Intelligence in Action has been running at Siemens Healthineers since 2023. 111 participants have completed the programme in six cohorts, and the Net Promoter Score has ranged between 57 and 100 across all cohorts (the NPS runs from -100 to +100, and any score above 0 already counts as good; above 50 is considered excellent). These figures come from participant feedback at the end of the eight weeks, not from a controlled study. They show how the programme is experienced, and they're one reason a company keeps booking it year after year. More about the results and the structure is on the programme page.

From practice, I can add what I keep noticing in trainings that work. Change needs repetition over several weeks, because new response patterns have to be practised before they're available under pressure. It needs real situations from your own working life rather than constructed case studies. And it needs a small group where people feel safe enough to say openly what's going on inside them. That's why our programme runs for eight weeks and happens live.

6. What this means for your team

From the data, I draw four concrete conclusions for organisations in the DACH region.

Emotional intelligence belongs in every AI rollout

If only 6% of leaders are making progress in designing how humans and AI work together (Deloitte), the reason is rarely the technology. It's uncertainty, resistance and unspoken worries. So build room for these emotions into every AI project, from the start.

Practice over weeks has more effect than a single day

The research shows emotional intelligence can be trained. In my experience, it needs time and repetition in everyday work to get there. If the budget is tight, a longer format for a smaller group is usually the better investment than a big one-off day for everyone.

Measure what you change

In a year when proof is required, every training needs a before-and-after measurement, even a simple one: a self-assessment of stress and how people handle conflict at the start and at the end, plus a Net Promoter Score. That lets you show impact when the next budget is negotiated.

Start with the people who lead

According to Gallup, managers' engagement is falling faster than their teams'. Yet managers set the emotional temperature of an entire team. If you want the biggest effect from a limited budget, start with the people who lead others.

One last thought

The numbers in this article are useful for justifying a budget or convincing a leadership team. What keeps me in this work, though, isn't in any report. It's the moments in a session when someone says out loud, for the first time, what has been weighing on them for months, and the others in the room simply listen. AI can now write us the perfect wording for a difficult feedback conversation. The conversation itself, with everything that happens in the body along the way, stays with us.

That's exactly why I believe emotional intelligence will matter more over the next few years: as a capacity we practise every day, in the meeting, in the feedback conversation, and in the seconds before it.

P.S. If you'd like to build emotional intelligence as a practice in your team, take a look at our 8-week programme Emotional Intelligence in Action. You can book a first conversation at cal.com/sophie-bachmann/discovery-call.

All sources

Frequently asked questions