For thousands of years, we’ve told stories about the genie in the bottle. Aladdin is probably the definitive mainstream version. The story of the impossible gift, the three wishes, where almost anything you could possibly imagine suddenly becomes within reach.

There are many versions of the genie-in-a-bottle story, but they’re all basically a giant thought experiment about power, unintended consequences, greed, and the fact that getting exactly what you ask for is often not the same as getting what you actually want. But in most of those stories, the hardest part isn’t getting the genie to grant the wish. It’s knowing what to ask for. We’re pretty good at knowing what we don’t have, and what we wish were different. We’re not nearly as good at figuring out what we actually want when the usual limits are taken away. More importantly, we’re not very good at anticipating what new problems appear once one hurdle has been removed. The genie was never really the interesting part. The interesting part was what happened to us when the genie showed up.

In a business environment, having access to AI tools like Claude, Gemini, ChatGPT, etc. has, for many, become a similar conundrum. Suddenly having access to powerful AI tools is like awakening the genie and being confronted with figuring out what to wish for. Luckily, we have more than three wishes in this scenario. 

We got more capacity, not more judgment

The first thing that happens when employees gain access to an AI tool is that it gives them a huge increase in capacity to produce cognitive output without necessarily increasing their capacity to decide what output is worth producing. 

Imagine a manager suddenly gets 10 competent-but-unsupervised junior employees.

If the manager has:

  • a clear objective,
  • good taste,
  • good judgment,
  • a backlog of valuable projects,
  • the ability to delegate,
  • and a way to evaluate results,

then this is fantastic.

But if the manager doesn’t know what the organization should actually be doing, the 10 employees don’t solve the fundamental problem. They create 10 times as much activity. AI is similar.

The scarce resource moves from:

“Can we produce this?”

to:

“Should we produce this?”

And that’s a much harder question to answer.

When removing friction creates more noise

In the latter example, we now enter into a really interesting failure mode: AI doesn’t just increase throughput; it can increase the throughput of nonsense.

Before AI, a person might have an idea and encounter friction:

“That’s probably not worth doing. It would take me three hours and requires others to be involved.”

That friction is surprisingly valuable, but AI now removes it:

“It’ll take me 30 seconds, and I can do it on my own. Let’s see what happens.”

So now people can pursue vastly more marginal ideas, where each one seems harmless on its own, but collectively, you get: more emails, more presentations, more documents, more meetings, more proposals, more analysis, more code, more marketing, more Slack messages, and maybe worst of all, more “thought leadership”.

This is what I’d call cognitive inflation.

The amount of stuff being produced goes up faster than the amount of stuff that is actually useful, which often creates a nasty feedback loop:

AI → more output → more things to process → more attention required → less attention available → lower-quality judgment → more AI-generated output to compensate.

“AI has enabled an enormous expansion in human execution capacity, but our ability to determine what is worth executing has not expanded proportionally.”

Peter Nilsson

There’s another consequence too. Every organization has people who consistently create value, and others who unintentionally create work. Historically, a lot of that extra work was filtered out by the natural friction of limited time, limited resources, rigid processes, or simply having to convince someone else before moving forward. AI removes much of that friction. The challenge for managers isn’t that AI amplifies people. It’s that it amplifies both the people who create value and the people who create work.

AI is still an extraordinary accelerator

Now, I don’t actually think the problem is that most people don’t know what to do with the capacity. Humans have an enormous number of things they would like to do but can’t because of labor constraints, and this is where AI tools become incredibly powerful. Looking back in a few years, I am confident we’ll look at AI as one of the greatest accelerators of innovation because it removed a lot of friction that historically has killed brilliant ideas by smart but under-resourced people.

Consider someone who has:

  • a business idea but can’t build the prototype
  • a small business but can’t afford a marketing department
  • a researcher who can’t analyze all the literature
  • a teacher who can’t personalize instruction for 100 students
  • a nonprofit with 30 projects it can’t staff
  • a programmer who has 50 improvements they’d like to make
  • a person who wants to learn a difficult subject but can’t afford a tutor

For these people, AI isn’t primarily “more employees”; it’s a dramatic removal of bottlenecks, and that’s potentially enormous.

Capacity only creates value when it has direction

But AI doesn’t automatically create value from additional capacity. It magnifies the relationship between capacity and direction. So if you’re already pointed at something valuable, AI can be an extraordinary accelerator of your time to value. On the other hand, if you’re poorly directed, AI will make you extraordinarily productive at being poorly directed. Having employees, human or otherwise, isn’t primarily valuable because employees can do work. It’s valuable because someone can manage a production system where added capacity increases throughput and business outcomes.

Whether you manage only your own time or you’re guiding a group of resources, a good manager knows:

  • What are we trying to accomplish?
  • What should we not do?
  • Which tasks should be delegated?
  • What does good work look like?
  • How do we check it?
  • How do the pieces fit together?
  • When should we stop?

The genie was never the point

And perhaps, that is the real lesson of the genie-in-the-bottle stories. Where the genie was never particularly interesting because it could grant wishes, it was interesting because it forced us to confront what we would do if our life constraints suddenly disappeared. AI is giving us a version of that genie, and the temptation is to focus on how much more it can help us produce. But the more important question is what we choose to produce when production itself becomes cheap. The organizations and individuals who benefit most won’t necessarily be the ones who make the most wishes, or even the ones who have the most powerful genie. They’ll be the ones who develop the judgment to know which wishes are worth making, which ones aren’t, and when to stop asking.

We may have finally awakened the genie, but the genie was never the scarce resource.

Knowing what to wish for was.