Productive Obesity

“In some ways this will be a new era for humanity, but in others it’s just a continuation of historical trends. As recently as 200 years ago, 90% of people were farmers growing food to survive. Advances in technology have steadily freed much of humanity to focus less on subsistence and more on the pursuits we choose.”

— Mark Zuckerberg

Mark Zuckerberg wrote this in his letter on superintelligence, and I find the parallel he draws between agriculture and AI to be more revealing than he probably intended.

Historically, humans worked to produce the food they consumed. The effort of growing, harvesting, and cooking a meal burned a significant portion of the calories that meal provided. For the first time in history, the inverse is the case. We drive to pick up food, or get it delivered to our door. We eat with little to no energy expenditure and then focus on how we’re going to burn the calories off afterwards through exercise.

The Calories of Production

The advancement in agriculture shifted our relationship with food from scarcity to abundance, and that abundance created a problem that was previously impossible: obesity. Before any of these advancements, the effort required to grow, harvest, produce, and cook the food necessary to become obese would burn enough calories in the process that it could not happen. Through efficiency, we came to realize that the bottleneck had been our bodies themselves.

It is clear that the same shift is happening with AI and productivity. The inefficiency that came with work before AI allowed us to be deliberate. A programmer could not code an entire codebase within an hour, much less a few hundred lines in a day. Every task had a cost of expenditure that made the programmer ensure he would not have to rewrite that line of code again. The same goes for any other professional knowledge work. The cost of making mistakes, of wasting your time and effort, was so high that a standard of precision was demanded.

The Zero-Cost Bottleneck

What happens when that cost goes to zero?

AI has allowed us to become productive beyond our wildest dreams, and because of that we encounter a problem that historically was not possible. When you work at a slow pace, and each task requires weeks if not months to achieve, you get time to reconsider. You get incentives to reduce your opportunity cost, to reflect on whether this is valuable to you, whether it will actually get you to your outcomes. As you move faster, that time to process and reflect on the value of your work shrinks. It was never previously possible to work faster than your decision-making could go. That is now the case.

The Illusion of Labor

This is no clearer to me than with the vibe coding trend. I’ve watched agencies spin up entire web applications on tools like Lovable in a matter of hours. One agency built a client portal for their marketing services company, shipped it in a weekend, and congratulated themselves on the speed. Within two months the client’s churn went up. Customers complained the portal was janky and unreliable. The agency had moved so fast that nobody stopped to ask whether the thing they were building would actually retain customers or push them away.

Think about it: how many times has a portfolio company shown up to a board meeting with a new AI dashboard thingy that nobody on the operations team actually uses? While everybody tries their hand at the new AI tool of the day, through their obsession over whether they can do something, it is lost on them whether they should do that something. The concept of calories only became important once food was in abundance. Now we live in an age of productivity abundance, with countries of geniuses at our disposal and every influencer entrepreneur running a fifty-‘person’ company from their laptop. It has become extremely easy to fool yourself into believing you’ve done a day’s worth of work without having actually achieved anything productive.

Designing the Experiment

The fix is not to slow down. That would defeat the purpose. The fix is to treat AI implementation much like a scientific experiment. Your hypothesis is that the AI tool is going to make your business more efficient, but that is not enough. You need a single business outcome as your North Star. One KPI. Not five dashboards or a suite of automations. No complete digital transformation. One number that needs to move, by how much, and by when. Hold the AI factor to a verifiable result with controlled variables, or you risk losing yourself in the maze of productivity, taking lefts and rights that get you nowhere.

As the market rushes to usher in the new age of AI, selling chatbots and automations to every business that gets on a sales call, I believe we need a calorie-conscious perspective to our work. The purpose of AI in a business should not extend beyond the purpose of an employee: to create value for the company and contribute to the outcomes that drive that value. Everything else is an illusion of labor.