The recent Harvard Business Review piece, ‘How People Are Really Using Gen AI in 2025’, reveals something I’ve suspected for a while - the line between AI and human ‘soft skills’ is blurring faster than many predicted.

The Shifting AI Landscape

Take a look at the fascinating changes from 2024 to 2025 in the Top 10 use cases:

*(image copyright Harvard Business Review)*

The End of ‘Humans Will Always Be Better at Soft Skills’

What stands out immediately in the graphic above?

To me, it’s the dominance of ‘personal and professional support’ use cases, with ‘therapy/companionship’ at number one, and the new use case of ‘find purpose’ at number three.

We’ve all heard the reassuring mantra, echoed in so many articles about AI in the workplace, that AI will excel at many things, but humans will continue to bring the unique soft skills of empathy, connection, and emotional intelligence.

But the data suggests that may be an attempt to cling to a comforting myth - or at least an only partial truth.

I think there’s been an element of wishful thinking to this idea for while. When I attended the AI for Good Summit all the way back in July 2023, researchers were already presenting findings from a telehealth project where AI ‘doctors’ were rated more highly than human physicians for bedside manner, empathy, and the ‘human skills’ of medical communication.

What we’re seeing in these HBR use case findings is simply the playing out of that reality at the mass user level.

Real People, Real AI Emotional Coaching

Think using AI for therapy sounds far-fetched? I know several leaders who are already doing exactly this in sophisticated and thoughtful ways:

The leader who is using AI to help unpick old and unhelpful thought patterns and belief systems, and intentionally identify his value system.
The leader who is using AI to process the emotional and psychological impact of major life change.
The leader who uses AI to prepare for difficult conversations, identifying his own blind spots and gaining perspective on his own emotions before going into an interaction with team members.

These aren’t replacing human relationships. But they are providing consistent, easily accessible and judgement-free spaces for reflection and growth that many find valuable.

The Self-Reinforcing Growth Loop

What’s particularly fascinating to me is what I see as the potential self-reinforcing loop being created:

AI Improves: AI continues to get better at soft skills, including therapy-related interactions.
Human Bar Raises: To outperform AI in ‘human skills’, leaders need to strengthen their self-awareness, emotional intelligence, differentiation, regulation, and empathy.
AI As Development Tool: AI provides an accessible mechanism for motivated humans to deliberately practice and strengthen these skills.
Both Continuing Improving: AI’s ability to demonstrate these ‘soft skills’ grows (including through all the data gained by humans using it as a development tool), pushing humans to strengthen their own skills further, continuing the cycle.

This isn’t just theoretical. One recent study showed people could not tell the difference between the written responses of ChatGPT-4o and expert therapists, and that the AI-generated content was rated highly by therapists and outperformed professionals.

What This Means for Leaders

The implications for leadership are profound. Sure, the very best humans at empathy, emotional awareness, and similar social skills will likely (or possibly?) always perform AI. But most of us, by pure statistics, aren’t the very best. Most of us still bring emotional blindspots, less-than-perfect self-regulation and empathy, and plenty of other weaknesses to our ‘human skills’.

These skills will indeed become even more crucial in leadership as AI use continues to expand. But we humans won’t automatically remain better at them - if we even are now.

Having a trained, expert psychologist or therapist (human and AI) helping us continue to grow our emotional health, self-awareness, and interpersonal toolkit will become as non-negotiable to ongoing leadership effectiveness as mastering productivity skills.

Ethical Considerations and Limitations

Of course, all of this raises important questions. AI lacks actual emotional experience and any spiritual dimension, ethical frameworks are still developing and often sorely lacking, data privacy concerns are real, and the data sets on which LLMs are trained can often perpetuate a range of biases. An AI functioning to provide therapeutic services won’t pick up on subtle physical cues (at least not yet, but video inputs are already rolled out), and is probably not appropriate for serious mental health conditions.

Yet for many leaders seeking growth, reflection, and skills development in a safe space, the benefits are often outweighing these concerns.

Moving Forward

As you think about your own leadership development, consider:

How might AI tools grow and strengthen the ‘human skills’ that differentiate great leaders?
What emotional or interpersonal challenges could benefit from consistent, judgement-free reflection or processing practices?
Where might you be holding onto assumptions about AI capabilities? (and ‘unique’ human capabilities) that need updating?

How have you found yourself turning to AI for personal processing, growth, or development? And how might you continue to explore more possibilities?

The Changing Face of AI and ‘Human Skills’