The skills required to do your job will change by the time you finish reading this sentence.

Kidding. But they are changing fast, and the rate is expected to increase. Since 2015, the skills needed for many jobs have shifted by 25%, according to data from LinkedIn. That number is expected to hit 65% by 2030, largely driven by technologies like AI.

How are we supposed to keep up? Kian Katanforoosh, CEO of skills intelligence platform Workera, says that people need a strong foundation of “durable skills,” like communication and statistics, that allow them to quickly and continually acquire new “perishable skills,” abilities that are less evergreen but that keep them at the cutting edge.

We spoke with Katanforoosh about this distinction between durable and perishable skills.. Here are highlights from that conversation, edited for length and clarity:

Can you explain the distinction between durable skills and perishable skills?

A skill is durable if its half-life is above five years. That’s my mark. Communication, for example, is above five years. AI is moving so fast that if you are lacking durable skills, you will not be able to learn the next perishable skill that is going to appear. If I want to learn the new version of Python, what’s going to help me the most is not that I know the previous version of Python only. It’s that I understand how software is built, I understand how algorithms work, I understand how data is processed. Those are skills that have a longer half-life.

One of the topics that CEOs were talking about at [the World Economic Forum annual meeting in] Davos was, ‘We need to move from prototyping to production. Since ChatGPT launched, we have put together so many prototypes of [generative] AI.’ You see your teams giving you demos, but for some reason it doesn’t get to production. One of the reasons you’re observing that is because to get to production, you need another level of skills that comes from some foundational areas. Oftentimes the new generation that has been learning AI has skipped them just because the trend is at the latest language model. So it’s hard to expect a team to do proper AI production when they don’t have the foundations of linear algebra, of mathematics, of algorithmic coding, of software engineering, of cloud computing, et cetera.

Sign up for Charter's newsletter to get the handbook for the future of work delivered to your inbox.

What about for less technical roles?

It’s the same thing. It’s just that durable skills are defined differently. Algorithmic coding is a good durable skill for a technical person. For a non-technical person, I would say having a good grasp of the ethics of AI is a durable skill for a business practitioner. You’re going to reach a point where you have to make a complex decision, and being able to go back to the ethics of the technology will allow you to sprint faster to make the decision faster and better.

I’ve made this distinction between sprinting and marathon—they’re learning styles. I think everybody needs to have both mindsets. My marathon is developing as an executive. I want to continue to become a better people manager. I want to become more mindful in my communication. I want to also master a broad set of technologies between data, cloud, AI, so that whenever something comes up, I can sprint. The sprinting mindset is more like, if I have a problem that arises—let’s say there’s a new technology that appears in the market—I want to be able to sprint toward it. When ChatGPT was released, I wanted to be able to know everything about it. When RAG [retrieval-augmented generation] became popular earlier last year, I wanted to sprint and understand it very deeply.

How do you gauge a job candidate’s learning style?

At the end of an interview, I ask, ‘What are three things that you learned in the last 90 days?’ Then I dig into it to see if it’s just anecdotal or if they’ve actually learned it. I may ask a few targeted questions to verify that they learned it. Second, I ask an open-ended question, which is, ‘What is your learning strategy?’ or ‘What is your learning architecture?’ That question usually triggers self-reflection from the candidate: What are the resources that I put around myself to make sure I’m continuously learning?

I can clearly see candidates who have put together an actual framework around them. ‘I have contracted with a coach who I talk to every two months, and these are the things I cover with them.’ ‘I have a subscription on Audible, and I set myself the goal of reading one audio book every month.’ Or, ‘I have this research reading group that I participate in on a weekly basis that I initiated.’ People would have one, two, three, sometimes more of these learning strateg[ies] around them, that allow them to continuously learn. That question triggers some of that.

Read more from Charter

The handbook for the future of work, delivered to your inbox.