We host a podcast, Strategy Skills, where we interview senior thinkers across business, strategy, and increasingly, artificial intelligence. There is one question I have been asking the guests for some time now.

The question is some version of this. Given everything you have just told me about the trajectory of this technology, what skills should serious operators and executives develop?

The answer I get most often is something along the lines of “Learn to work with AI.” Get fluent with the tools. Build the workflows. Understand what the systems can do and integrate them into how your team operates. It is not bad advice. It is also incomplete, and I will explain why.

Notice something else first. The advice from my guests has been roughly stable across interviews. The public messaging from the AI industry itself has not.

Shortly after ChatGPT came out, some leading AI companies and executives publicly emphasized highly disruptive, replacement-level capabilities. In fact, we could see at least one leading company having a stated mission of the development of systems that would outperform humans across most of what people currently do for a living. The underlying claim was that human labor was on its way to obsolescence in many areas where it is currently deployed, and society would need to figure out what humans should be doing next.

That message has been rapidly walked back, though. Public messaging from some of the most visible AI leaders has shifted more recently toward augmentation, productivity, and partnership with humans. And the shift happened surprisingly quickly.

What is driving it? American polling on AI is showing concerning signs. Recent Pew data shows the share of Americans who believe AI will have a negative impact on the country has risen to roughly half, with the strongest opposition coming from independents. Multiple US senators have proposed legislation to expand federal involvement in AI research, oversight, and lab infrastructure. The current administration is reportedly considering a formal government review process for new AI models before public release. Industry leaders who six months ago would not have entertained the idea of regulation are now publicly discussing what kind of partnership with the government they would accept. The pitch was adjusted.

I am not here to evaluate the sincerity of any individual executive. People in positions of influence sometimes change their views. They also sometimes change their messaging. The two are different and not always distinguishable from the outside.

The right question is not whether the new sales pitch is sincere. The right answer is not, by itself, “learn to work with AI.” The right question is whether the pivot in industry messaging changes anything about what serious operators, executives, and professionals should already be doing.

It does not. And the reason involves something I have not heard a single AI guest articulate clearly on our podcast.

The first thing

I have lived through one inevitability narrative already. I grew up in the Soviet Union. I was raised inside a system that told its citizens, with full institutional confidence, that the direction of the country had been settled. The media repeated it. The schools repeated it. All of the people I knew growing up believed it.

Then the Soviet Union collapsed. Not gradually. The framework that had structured the lives of three generations turned out to have been a story that dissipated in record time. The moment the system could no longer enforce the story, the story stopped being true.

I do not equate Soviet ideology with the public messaging of AI companies. They are not the same thing. But I learned something growing up that has stayed with me forever. When you hear an institution insist that a particular outcome is inevitable, the first thing to check is whether the institution making the claim benefits from your believing it. The second thing to check is whether the people on the inside actually agree with one another.

In the AI industry, the answer to the first question is yes. The answer to the second question, given how quickly the messaging has changed, is at least unclear.

This does not tell you that any particular prediction is wrong. It tells you that your career and your team should not be organized around someone else’s predictions about the future. Your own judgment, your own opinion, should be your guiding star.

The second thing

The professionals who will be valuable in five years are not the ones AI happens to replace. They are not even the ones AI happens to augment, because that is a temporary advantage. They are the ones doing the kind of work that remains much harder to automate reliably.

This distinction has been largely missing from public discussion. It is the most important distinction in this article.

If your work consists of accumulating structured information from various sources, sorting it, summarizing it, formatting it, and drafting it from templates, then your work was always going to be automated. AI is just the thing that arrived to do it. Whether AI will create new tasks for you in its place is an interesting question, but it is not the most important one for you, because I don’t think you want that type of work anyway. The more important question is whether you have built or can build a career where you can deliver meaningfully higher value versus AI and versus your peers, because being average has never been more dangerous, especially in certain professions where humans are being augmented and then replaced by AI first.

If your career is built on a powerful skill set far superior to that of an average performer, AI is not a threat to you. In fact, you can use it as leverage.

Michael and I have spent decades in consulting, corporate, and working with eminent clients, and we know as much as anybody that the standard analytical work has always been valuable to a junior professional. Building models, structuring problems, producing slides, running benchmarks. None of it has been the source of a senior professional’s actual advantage. The senior professional’s advantage has always been the ability to see what is hiding in plain sight, but others don’t see, to enter a complicated situation, see what matters in it, decide which problem is the one worth zeroing in on, communicate that judgment in a way that connects with a decision-maker, and respectfully push them to act and drive the project to successful completion.

None of that has changed. AI compresses the time required to produce the underlying analyses. It can generate plausible ideas, but it does not reliably identify which problems matter most or produce solutions that are both contextually correct and executable in complex environments. It also cannot stand for you in front of a senior client or a board of directors and be clearly trustworthy. The professionals who already had solid business judgment and powerful and rare problem-solving skills now have more time to apply them. The professionals who confuse producing the material with having superior business judgment are exposed.

This is uncomfortable for some to hear, but it doesn’t change the fact that it is true.

The third thing

There is a category of work that almost everyone in the AI discussion now agrees will remain valuable, regardless of which prediction about jobs turns out correct. The category is sometimes described as the human element, sometimes as relational work, sometimes as the work that requires presence.

The framing is roughly right. The language is not concrete enough.

The work that will remain valuable is not “the human touch” in any sentimental sense. It is the work of thinking clearly under pressure, holding the attention, building trust-based relationships, being able to accurately evaluate the situation in front of you, communicating with authority, seeing insights even AI will not be able to see, and inspiring confidence in others that the decision you are suggesting is the right one.

These are among the most valuable and hardest-to-develop skills in business. They were rare before AI, and they are becoming rarer because the training ground for junior people to build these skills is shrinking by the day. The required conditions that produce these skills cannot be underestimated. Decades of operating experience. Someone sitting there and teaching you or giving you right-hand experience. A sort of apprenticeship. Being there in tough situations and navigating those. Seeing how someone brilliant is navigating tough situations. The willingness to be the person who carries the weight of a hard decision. None of these conditions are affected by AI in either direction, other than that junior people will likely have fewer and fewer opportunities to develop those skills. These skills are developed inside human professional lives over time, over decades, and only if you are lucky enough for someone to teach you or brilliant enough to figure out key principles by trial and error.

If you are looking for a defensible position over the next decade, this is where it is. Not in beating the machine at the work it was designed to do faster than humans. In doing the work the machine struggles to do reliably today and is likely to remain constrained in for the foreseeable future. Developing these skills was hard before AI, and now opportunities are fewer and farther between. In fact, StrategyTraining.com was initially created to offer an opportunity of a sort of apprenticeship with our coaches. For example, Michael obviously would not have time to train thousands of junior partners one by one. In fact, that is why we keep executive coaching groups so small. But we did create an opportunity to be Michael’s apprentice remotely, via Insider or Legacy membership.

The fourth thing

The new industry messaging by some AI players talks about AI as a tool that gives humans superpowers. And there is truth to it, but with a caveat. You need to be a sort of superhuman (in terms of skills, drive, determination) to make AI something that gives you superpowers vs. your competitors or peers. Most professionals now use AI as an assistant, and many have agents deployed.

Building AI that augments human judgment is a different design choice than building AI that mimics human work. The first compounds the value of human capability. The second tries to copy it and is only useful when the copy is cheap. Most of the capital in the AI industry has flowed into the second category.

We are internally building one of the first kind. The experience of building it has clarified something for me. The augmentation framing is correct in principle and rare in practice because building AI that helps a human think more expansively and deeply than they could on their own requires that you understand, deeply, how the best humans actually think. That is a much narrower and more demanding starting point than imitating human output.

Why I am writing this, and who Michael and I are

I have worked alongside Michael for many years. Michael spent his career as a strategy consultant working directly with CEOs of major companies, sat on the operating side of large organizations, and has spent the last fifteen years training senior strategy and operating professionals through the work we do at FIRMSconsulting and StrategyTraining.com, on top of other work we are doing, such as full-blown consulting studies. The professionals who come to us are senior consultants, partners at major and boutique firms, executives running P&Ls, and operators who have already had successful careers and want to keep having them as the field changes and the world is more uncertain than ever.

For three years running, I have been recognized as one of the top thirty management thinkers worldwide, but it is really Michael and me because the books reflect not only mine but also Michael’s deep expertise in strategy, risk, operations, and implementation.

I am writing this article because I have watched the AI industry pivot, in a matter of months, to a version of the same advice Michael and I have been giving senior professionals for over a decade. We did not need the polling to collapse. We did not need a regulatory threat. We have been telling our clients since long before AI became a public concern that the work worth investing in is the work that does not depend on which technology cycle is in fashion. Build the business judgment. Build the executive presence. Build the ability to walk into a difficult engagement and be able to get it back on track and beyond. The professionals who built their careers on this principle are the ones doing well now, regardless of what their company has decided to do with AI. The professionals who built their careers on producing analytical output and who are unable to think for themselves (thinking for yourself is what we teach) but constantly rely on someone else to tell them what to believe are the ones secretly looking for another path, regardless of how good they were at the analytical output.

This is a difficult thing to commit to. By the time the industry confirms that the principle was correct, the professionals who acted on it five or ten years ago have a head start. The professionals who waited for the consensus to form are now learning that the consensus does not, by itself, develop their capability for them.

If you have read my writing before, you already know this is what we teach. If you are new to our work, this is the question Michael and I have spent our careers answering for senior people who came to us specifically because they wanted a perspective that would not change with the next cycle.

The answer has not depended on the cycle. It has been consistent because the underlying truth has been consistent.

What the answer comes back to

When I ask my podcast guests, Harvard and Stanford professors, CEOs of Fortune 500 companies, McKinsey and BCG partners, what skills serious operators should be developing and telling their teams to develop, the standard answer is to learn to work with AI. That is necessary. It is not enough.

The expertise that allows you to create outsized value and which does not depend on which prediction turns out to be correct is the expertise worth investing in. Develop your and your team’s business judgment, executive presence and communication skills, ability to build trust-based relationships, and superior ability to solve the toughest problems. Invest in your and their ability to think clearly under pressure. Invest in your and their ability to communicate in a way that holds attention and inspires confidence. Invest in your and their ability to enter a difficult situation and be the one capable of fixing the issue. These capabilities matter regardless of what the technology advancements we see next. They are also the capabilities that the technology cannot, by its nature, take from you. It goes without saying that you want to use the technology available. That is table stakes. That is not enough.

The shift in messaging from the AI industry is a useful occasion to step back from other people’s voices and opinions. Behind the messaging, on either side of the pivot, the things that make you valuable did not change. It will not change in five years. Most likely it will not change in twenty.

The professionals and organizations that build for that will be the ones still standing when the next narrative arrives, and the one after that. The professionals who keep waiting for an authoritative source to tell them what to do and how to think will keep being at least a cycle behind. They will adjust, and adjust again, and never catch up. Investing in yourself, as Warren Buffett said, is the best investment you can possibly make. Don’t be fooled by AI being able to think for you. Be the one who is a far superior professional able to deliver outsized value that others can’t. AI is trained on large distributions of existing work, which means it tends to reproduce patterns of typical performance rather than exceptional judgment. In other words, we could say that AI is trained on a lot of average work. And very few people in our community are ordinary. And, certainly, no one is average who got to the end of this prolonged thought piece.

Kris Safarova

CEO, FIRMSconsulting.com & StrategyTraining.com