How to keep your procurement skills sharp in the AI era

My phone died on vacation this summer in Venice, a city I did not know well, with three generations of my family waiting on me for directions.

“I know,” I said. "We can use a map!"

I quickly discovered that I am actually not as good at reading and navigating with maps as I used to be. This is an effect known as ‘skill atrophy,’ and I couldn’t shake the feeling that the same thing is happening right now with AI at work.

Three generations and one dead phone battery

Up until the phone unceremoniously ran out of battery we had navigated that entire trip by GPS, the way most people do in their day-to-day. As you know, all you have to do is type an address into your phone and your Maps app finds the fastest path, be it walking or driving. But as our phone battery went, so did the maps.

To use a paper map was a genuinely novel concept for my son, who has never needed to in his life. For him, this was a new challenge. For my mom, who did not grow up with cell phones, it seemed perfectly normal. What I found interesting was how this clash of generational skills affected me.

I grew up navigating long road trips with a giant book of maps. but at this point this is merely embedded in muscle memory. Could I still do it? Yes! Was it as easy as I remembered? It was not. My ability to navigate with a map instead of GPS had clearly weakened, as if the skill had atrophied over time.

The same thing happened when we tried to calculate a tip at a restaurant. I also did not grow up with a calculator in my pocket, so doing math in one’s head was simply expected. Same with phone numbers and birthdays; before cell phones, I memorized my friends' numbers so I could reach them from anywhere. Once my phone remembered them for me, my ability to do so diminished year after year.

Why AI skill atrophy matters at work

AI has become a necessary tool for productivity. In Zip's State of AI in Spend 2026 survey of 1,050 procurement, finance, IT, and operations leaders, 62% said they use AI tools multiple times per day. AI is well past the experiment phase; it is quite literally how work gets done today. And in some cases, these tools can cause you to lose skills you previously had.

Take the common use case of writing an executive summary of a project. 

Before AI, this was a manual task that required real thought. You had to know the high-level story of the project, understand the audience so you could tailor the message, and condense everything into a format that got to the key issues.

Now an executive summary is a click away. That is a huge time saver to be sure, and the productivity gains are real; 89% of leaders in the same survey report a net-positive productivity effect from AI, even after accounting for the time spent reviewing and correcting its output. Sounds like a solid win.

But can it lead to skill atrophy? Possibly. If I no longer regularly work through those questions myself, I may lose some of that ability over time. Most professionals are not worried about this yet. Half of surveyed leaders say AI is augmenting their skills while their skill set stays largely the same, and 32% say AI is helping them develop new skills faster. But 12% report a mixed effect: AI helps in some areas while they lose sharpness in others. That 12% is describing exactly what I felt holding that paper map.

So what we have to ask ourselves now is does it matter?

The risk of over-reliance on AI tools

I would argue it matters. Going back to my examples, technology often fails. What if I lose access to my AI tools at the moment I need to produce a summary? Am I so reliant on AI that I can no longer do the task at all, even less efficiently?

This is not a hypothetical. When leaders were asked what would happen if their primary AI tool were unavailable for a single workday, 72% said they would feel at least a moderate hit to their productivity, and 29% said the impact would be significant or severe. One workday.

And sometimes you do not have the tool in the moment that the question arises. Have you ever been cornered in a meeting and asked to summarize a project you had not prepared for? Can you respond effectively without saying you need to go back to your desk and ask AI for help?

Why fact-checking AI output is the skill to build

There is a second side to this, which is that many people who have not embraced AI at work hold back because they do not trust what the tool produces. That skepticism keeps them from real productivity gains. What often gets overlooked is the opposite side of the coin. What happens when you lose the underlying skills and rely on AI outputs with no ability to question the result?

A good example is the navigation app Waze. It gives you the optimal route based on live traffic. Have you ever blindly followed Waze only to find its "optimal" route asking you to make an unprotected left across multiple lanes of rush-hour traffic? Follow those directions without question and you may sit at that intersection for a very long time.

The same applies to AI. If you follow its output blindly because you have lost the ability to interrogate it, you may end up on a longer or incorrect path. Leaders seem to sense this.

When asked which capability their teams most need to develop for an AI-driven environment, the top answer, at 49%, was evaluating and fact-checking AI-generated analysis. The skill of the next few years is the discernment and base-level expertise to know whether the answer AI gives is actually correct.

Why AI over-reliance is a business continuity risk

I spent the majority of my career as a procurement practitioner and leader. I know business is rarely this simple. Reliance on technology is normal; over-reliance can be catastrophic. In the earlier days of the software industry, procurement and security professionals focused heavily on disaster recovery. What happens to the business if this piece of technology stops working? How do we stay afloat?

AI may not require a disaster recovery plan in the traditional sense, but the same discipline applies. And most organizations are not there: only 31% of leaders say they are very confident in their business continuity if a critical AI system failed. Over-reliance and skill erosion are a continuity risk of their own, just one that shows up in people instead of servers.

None of this argues for using AI less; it still has tremendous upside. Be deliberate about which skills you keep sharp, so that when the battery dies, you can still read the map.

The data in this article comes from Zip's State of AI in Spend 2026, a survey of 1,050 procurement, finance, IT, and operations leaders fielded April to May 2026. Download the full report.