There’s a comforting story going around that as AI handles more of the technical work, we’ll all retreat to the “human” skills and everything will be fine. The first half of that is true. The second half skips over something inconvenient, which is that the human skills everyone’s pointing to are the hardest ones to actually build.
The skills that are quietly winning
Start with what the data shows. HBR researchers looked at 70 million job transitions and found that foundational skills, things like collaboration and adaptability, did a better job predicting who learned faster and earned more than narrow technical specialization did. The specialist skill got you in the door. The foundational skill was what moved you up.
That’s a shift from how a lot of us were told to think about careers. The advice used to be to go deep on one technical thing and become irreplaceable. The finding here is closer to the opposite: the people who adapt across situations and work well with others have the more durable advantage, and it compounds.
PwC’s work adds a concrete reason why. Looking at the new tasks piling onto AI-exposed jobs, they found those tasks are about 2.5 times more likely to demand empathy, judgment, and creativity than the tasks the AI took over. So as the routine work drains out of a role, what floods in to replace it leans hard on the human stuff. The jobs aren’t disappearing. They’re getting more human, and more demanding in a way that doesn’t show up on a skills certificate.
The catch the advice skips over
There’s a catch the “just pivot to soft skills” advice glosses over. These skills are genuinely hard to build from scratch, and you can’t cram empathy the way you can cram for a certification exam.
Lynda Gratton of London Business School put it about as directly as I’ve seen, writing in HBR that the capabilities leaders value most in their employees are also the hardest to develop: discernment, intuition, moral reasoning, and, above all, empathy. Read that back and notice what it’s really saying. The most valuable skills and the hardest-to-build skills are the same skills. That’s not a coincidence. They’re valuable because they’re hard, and they’re hard because there’s no shortcut through them.
You build discernment by making calls and living with the results. You build empathy by sitting with people who see the world differently and actually listening, often when it’s uncomfortable. None of that fits in a course syllabus, which is precisely why it stays scarce even as everyone agrees it matters.
Why the timing makes this worse, and better
There’s a wrinkle worth sitting with. Gratton’s larger worry is that the “practice, frustration, honing craft” phase, the grind where these skills usually form, is exactly the part AI is good at engineering away. When the tool handles the tedious early reps, you skip the struggle that used to build your judgment as a side effect.
That’s a real trap. It’s also, if you’re paying attention, an opportunity. If most people let the tool absorb the hard, formative work, then the person who deliberately keeps doing some of it, who wrestles with the messy problem before reaching for the assistant, builds judgment the others are quietly losing. The scarcity works in your favor if you’re willing to do the unglamorous part.
How to actually build these
Vague advice like “improve your communication” is useless, so let me be concrete. Put yourself in situations that require the skill and can’t be faked. Volunteer for the cross-team project where nobody agrees on the goal. Take the customer conversation that’s gone sideways instead of routing it away. Ask for the feedback that stings, and then sit with it long enough to learn something.
The through-line is that these skills come from reps under real conditions, with a consequence attached and someone honest telling you how it landed. Reading about empathy does approximately nothing. Being the person who has to hold a room together when a project is on fire, and getting a little better at it each time, does a lot.
Feedback is the ingredient people skip, and it’s the one that matters most. You can put in a decade of reps and barely improve if nobody ever tells you how you’re coming across. Most of us are terrible judges of our own collaboration and communication, because we hear our intentions and everyone else hears our impact. The fastest way to grow these skills is to find people who’ll tell you the unflattering truth, and then to actually change based on it rather than defending yourself. That second part is where most people stall. It’s also, not coincidentally, a display of exactly the adaptability the data says employers are paying for.
Keep track of these moments as they happen, too. The hard conversation you navigated, the call you made with incomplete information, the time you changed your mind because someone pushed back well. That raw material is both how you notice you’re growing and, later, what you’ll draw on when someone asks you to prove it.
Where it gets tested
And someone will ask. Behavioral interviews exist mostly to probe for exactly these skills, since they’re the ones a resume can’t show. When an interviewer asks you to walk through a conflict you handled or a decision you’d make differently now, they’re not making conversation. They’re checking for judgment and empathy under a bit of pressure, and the answer either has lived texture or it doesn’t. That’s a skill of its own, and like the others, it responds to practice.
The honest read
The human skills are your edge, but not because they’re easy or because AI hands them to you by default. They’re your edge precisely because they’re hard, slow to build, and impossible to fake for long. That’s frustrating if you wanted a quick fix. It’s good news if you’re willing to do the work, because the difficulty is the moat. Anyone can prompt a model. Far fewer people can read a room, hold sound judgment when the answer isn’t clean, and bring others along, and those are the people who’ll be worth the most.