How to Prepare for the Next 5 Years

You will want to have read this framework guide

Hey, Alberto here! 👋 Each week, I publish long-form AI analysis covering culture, philosophy, and business for The Algorithmic Bridge. Paid subscribers also get Monday how-to guides and Friday news commentary. I publish occasional extra articles.

QUICK REMINDER:

I’m running a 50% discount offer on annual subscriptions that ends on June 5th June 12th (extended one week). Get yours now and enjoy the discount forever.

If you read the free edition and have been thinking about subscribing, this is the best time. I won’t do another offer anytime soon.

The Great Wave off Kanagawa by Katsushika Hokusai, ca. 1830

I. HOW TO DEAL WITH UNCERTAINTY

You don’t know what AI will do to the economy, to your job, or to the way you spend your days. That’s great news for our topic today: how to prepare for a long period of uncertainty.

AI uncertainty is not a temporary condition. It’s not that we’re waiting for more data, and then we’ll know. No, we won’t know. We never do, actually—AI just makes it weirder because it belongs to a class of phenomena that Nassim Taleb calls fat-tailed: the range of possible outcomes is so vast and the extremes so disproportionately impactful that the average outcome—the most likely AI scenarios—tells you almost nothing about what to do besides “proceed as usual.”

Planning for the average future (which is normally a good rule of thumb) doesn’t work here. A world where AI finds itself recursively self-improving by 2028 is too distinct from one where it collapses the economy by late 2026. The mistake we tend to make with fat-tailed risk in Taleb’s sense is always the same: we optimize for the average scenario and get destroyed by reality. Familiar examples of a fat tail are the 2008 financial crisis or COVID. Well, those were already quite bad, and yet, AI is the fattest of all because by definition, anything is in the realm of possibility.

So, how do you prepare for something you can’t predict? Two things. One, you need to know what to focus on today. Two, you need to know whether to focus on something else tomorrow. It’s that simple. Let’s see what that looks like in concrete terms.

II. WHAT TO FOCUS ON TODAY

Taleb’s answer to fat-tailed risk is the barbell strategy. You load both ends and leave the middle empty. One end is “maximum safety”: the things that are valuable regardless of what happens, the evergreen stuff. The other end is “maximum exposure to upside”: cheap bets that pay off handsomely if a particular future materializes, the timely stuff. The middle—moderate effort, moderate risk, moderate reward—is where you die. Don’t hedge.

Applied to the next five years: the safe end of the barbell is deep fundamentals that will be useful even if AI agents take over the entire workforce. Some ideas: Writing clearly, reasoning through ambiguous problems, persuasion, domain understanding, managing people, judgment and taste, etc. This is a non-exhaustive list. (You can show this post to ChatGPT and ask it to complete the list with the context of who you are and what you want from life.) Don’t forget that the intensity of your focus is inconsequential if the thing you focus on is the wrong one.

There are no shortcuts for this end of the barbell. These skills were as valuable in 2005 as they were in 1905 and as they will be in 2105, no matter what AI does. The reason I can be sure of this is that they pertain to the realm of human-to-human relationships. Our world is built on a set of principles—with the human at the very center—that you can govern with the right skills. These skills are the floor of civilization, and they should be yours.

On the other end of your barbell is aggressive AI-native experimentation. This is where I focus most of my practical guides because I take for granted that you’re already working on the “always important” stuff. I’m trying to emphasize that both types of skills are a must if you’re to be prepared for the next 5 years.

What does aggressive AI-native experimentation look like? Among others: familiarize yourself with the basic concepts and tools, use them daily, build complex workflows to challenge your theoretical understanding, ship things in public to gather feedback, develop a hands-on intuition for what models can and can’t do, etc. This is your lottery ticket. They’re cheap to acquire (time and predisposition) and enormous in value if AI reshapes your field.

The dangerous middle is passive familiarity. Here’s what that looks like: you read articles about AI like this one, but don’t do the work. You watch the news and panic. You’re aware of the industry announcements and the drama, but don’t know shit about the technology. This moderate, mostly passive engagement gives you the feeling of keeping up while building zero actual capability. Keep doing that, and you’ll be tragically unprotected. You could also fall into the dangerous middle if you start to consider evergreen skills useless in the AI age, leading you to forgo taste and judgment, domain knowledge, or overall cognitive competence. Don’t do that.

So the barbell is, in practice, a time split. Spend part of your time on skills that survive every scenario (maximum safety, evergreen). Spend another part getting your hands dirty with AI tools in ways that feel excessive relative to what your job currently requires (maximum exposure to upside, timely). Spend no time on the comfortable middle where you learn just enough to have a vague opinion. As Taleb would tell you: comfort under fat-tailed risk is risk of getting crushed by a fat tail.

III. WHETHER TO FOCUS ON SOMETHING ELSE TOMORROW

The barbell tells you what to do today, and that’s good. But what should you do with the information that arrives at you every day? How can you update once the future starts to materialize? How can you weigh more the evergreen if the AI bubble pops? How can you weigh the AI stuff more if job postings start to ask for hands-on agent experience? That’s what the barbell strategy lacks: It doesn’t tell you when you need to rebalance. For that, you need a different tool.