Welcome to this edition of "One Question"!

Seventeen years ago, Tim Ferriss wrote The 4-Hour Workweek. He talked about working smarter, automating the mundane, and freeing yourself from the grind. The book became the blueprint for people who didn't want to wait until retirement to live the life they wanted.

Today, AI can read your email, clean up your calendar, summarize meetings, draft slide decks, generate social posts, and make dinner reservations. That's just scratching the surface.

So is anyone working less?

Six years ago I started a mostly solo consulting business and scaled up through a trusted network of contractors. The structure gave me flexibility to work on different types of projects, but it also left me responsible for things that weren't fun, like securing business insurance, creating slide decks, and working through procurement processes. AI has made most of those easier. Yet, I'm working more, not less. So is everyone I asked. Why?

Three patterns keep showing up, and a fourth that contains them all.

Todd Henry, in Herding Tigers, writes about "ghost rules", invisible limits teams place on themselves for no good reason. A ghost rule is something a team does because it's always been done. No one questions it. No one can explain it.

A client of mine had a Friday "weekly readout" meeting. Every team leader spent 45 minutes preparing their read out. Then they’d all meet for an hour to share content that had already been circulated over email. When AI entered the picture, their first instinct was to use it to generate the readout faster. Nobody asked whether the meeting needed to exist at all. After looking at the value of the meeting, the team killed it and kept the AI generated readouts sent over email. Six hours a week came back, and nothing broke.

AI is great at making ghost rules easier to follow, but it’s a trap. The first question shouldn't be "Can AI do this faster?". It should be "Should this be happening at all?"

Leidy Klotz, in Subtract: The Untapped Science of Less, writes about "subtraction neglect", the consistent preference for solving problems by adding rather than taking away. Teams become averse to removal, leading to more meetings, tasks, slide decks or projects.

I see this everywhere with AI. A marketing team gets a tool that drafts social posts in seconds and starts posting twice as often, instead of asking how many posts they need. A finance team layers AI dashboards on top of dashboards, leaving teams with more to look at for the sake of having more to look at.

The rare team that asks "What can we stop doing?" before "What can we automate?" gets compounding returns. Most teams skip that question entirely.

Most AI rollouts I've watched go like this: a company picks a platform, hands it to everyone, says "be more efficient," and waits. Six months later, adoption is low and nobody can point to a clear win.

The companies getting real value test before they invest. They start with a need, a desired outcome and test the tools to see if and how they can achieve that outcome. Teams test multiple approaches. They measure and keep the best one. Not every AI model is good at every job. Some are better at research, some at data processing, some at writing. You can't know which fits where without putting them to the test on real use cases.

These three patterns feed into a fourth one I keep falling into myself - the feasibility trap.

My definition of "feasible" keeps shifting as new tools come out. Things I'd written off as impossoble or not worth the effort get a new lens.

Sometimes that's progress. Did I need a tool capable of spinning up slide decks in seconds? Absolutely! I hate making slide decks, but I do need them.

Sometimes it just means I'm searching for things AI can do instead of asking whether those things need doing. Did I need a custom project management tool that ingests transcripts of all my meetings and turns everything into tasks? Nope! It was just more work to deal with so I went back to my spreadsheet which was working just fine all along.

Companies fall into bigger versions of the feasibility trap, rolling out AI because the tools are impressive, then looking for problems to point them towards.

The feasibility trap thrives on ghost rules, feeds subtraction neglect and lowers the perceived value of experimentation.

The best analogy I can think of is moving from an old house to a new one. You can buy a bunch of new furniture that looks amazing in the show room. Then you get it into the new house and nothing fits quite right given the shape of the room. A tape measure and some blue painter’s tape on the floor would have easily solved that problem before making expensive furniture purchases.

hen it’s time to pack, you could box everything up and bring it to the new house because that’s what moving is, right?. Or, you could go through all that stuff and determine if it should even make the trip. That old Hawaiian shirt from a vacation 10 years ago? Does it need to be boxed up just so it can hang in a new closet for another 10 years?​

While I don’t see a shorter work week in anyone’s future, I do see a path to a beautiful marriage between humans and AI if we start by asking, “What do we hate doing?” and “Why are we doing it?”.

Expose the ghost rules. Reward subtraction. Encourage experimentation.

​​

If you enjoy this newsletter, please forward it to 3 friends or colleagues. They can sign up at: https://www.joelalley.com/newsletter​

​​-Joe