Human Skills in the Age of AI

I’ll start with a confession: the thing I use AI for the most, every single day, is help with English.

I work for an American company. My team is global. Every message, every report, every piece of feedback I write is in English. I speak it, I write it, but it’s not my first language. AI has become my daily partner for that. It helps me sound clearer, catch awkward phrasing, and express ideas the way I actually mean them.

But that’s just the beginning.

What AI actually does for me

AI helps me analyze data. I know how to do it myself, but AI does it faster, at scale, and without getting tired on row 487 of a spreadsheet. It helps me write reports, refine the tone of a customer response, and automate the kind of repetitive tasks that drain energy without adding value.

In customer support, where I work, the applications are even more concrete. AI can automate quality analysis of support tickets based on predefined criteria set by the team. It can scan hundreds of interactions and flag patterns: what’s the quality of the responses, are agents following the guidelines, where are the gaps. A manager can then take that broad analysis and zoom into specific cases for qualitative feedback and coaching.

AI is also great at processing large volumes of tickets at once, identifying trends, spotting recurring issues, and surfacing the cases that generate the most customer frustration. Work that would take a person days, it does in minutes.

So yes, AI is incredibly useful. But here’s the part that matters.

Where AI gets it wrong

AI lacks judgment. And in both of the examples above, that’s where it breaks down.

It can flag a ticket as “poorly handled” without understanding the full context. Maybe the agent made a judgment call based on a conversation that happened off-ticket. Maybe the customer was abusive and the agent de-escalated brilliantly, but the AI only read the words on the screen and scored it low.

It can identify an issue as “recurring” without understanding the nuance. Five tickets about the same topic might look like a trend, but three of them could be the same customer submitting duplicates, creating a false pattern.

I’ve seen AI misinterpret internal system tags. In one case, it read an automated Zendesk label that referred to an internal workflow and flagged it as a product issue customers were experiencing. It looked alarming in the report. It was completely wrong.

AI reads the surface. Humans read the room.

Will AI replace managers?

I don’t think AI will replace managers in their core function: managing people. But I do think it will make certain types of managers less necessary.

Here’s the distinction. If your job as a manager is mostly assigning tasks, tracking deadlines, and making sure things get done, AI will eventually take a big chunk of that. It will help individuals self-manage, automate workflows, and reduce the need for someone whose main role is keeping the machine running.

But if your job is developing people, building trust, navigating conflict, giving feedback that actually changes behavior, creating an environment where people want to do their best work, that’s a different story. AI isn’t even close to doing any of that.

The managers who should be worried are the ones who manage tasks but don’t manage people.

AI as a new layer

I think about AI the way I think about organizational layers.

At a previous company, the structure looked like this: CEO, CTO, Technology Director, Support Manager, Support Coordinator, N3 Support, N2 Support, N1 Support. Eight layers. The CTO didn’t have to think about a lot of things because the layers below him were handling it. His job was to operate at a different level, a strategic one, because the operational layers absorbed the rest.

AI is going to occupy some of those layers. Maybe all of one. Maybe parts of several. The question isn’t whether AI will take over certain functions. It will. The question is: what level are you going to operate at when it does?

This is where the concept of combined intelligence comes in. It’s not human versus AI. It’s human plus AI. The professionals who figure out how to think and execute in combination with AI will be the ones who thrive. The ones who try to compete with AI on speed, volume, or data processing will lose every time.

So what should you develop?

If someone asked me today, “what skills should I focus on that AI won’t solve for me?”, I’d say: the human ones.

Empathy. Communication. Transparency. Conflict resolution. The ability to read a room and know that something is off even when the data says everything is fine. The ability to give feedback that doesn’t just correct, but develops. The ability to build relationships that make people trust you enough to tell you the truth.

These aren’t soft skills. That label has always bothered me. There’s nothing soft about navigating a conflict between two team members, or telling someone their performance isn’t where it needs to be, or building trust across cultures and time zones.

These are the hard skills. They just don’t come with a certification.