tl;dr copy or download my 12-week cognitive fitness program (a free google sheet you can modify or use however you want). It’s mostly Claude-generated; a simple spreadsheet that I built for personal use.

AI’s deep work problem

As we adopt and leverage AI tools, our opportunities for deep focus diminish. While AI tools are often blamed for causing skill atrophy due to cognitive offloading, it’s plausible that a portion of that atrophy is due to lack of focus. These tools are marketed to help eliminate toil, but to what end? What are you skipping to? Where are you applying focus and deep work, and are these tools helping you or distracting you?

If you have a job that isn’t imminently getting automated away, chances are you can think of where you apply deep work and critical thought. One side-effect of AI adoption is that the time needed to apply this focus is swept away in a sea of pseudo-productivity. Most of the time in modern work environments, we find ourselves “babysitting” AI agents, juggling notifications, and bouncing text back-and-forth to chat bots until it reaches a tolerable level of cogency.

This is causing people to feel and act dumber. And without intervention, this dumbness is going to cascade into our collective output, amplified by our sycophantic AI-powered systems.

The intervention: cognitive fitness

Just as the dawn of sedentary knowledge work has spawned the quarter-trillion-dollar (and growing) fitness industry, I believe there will be a cognitive fitness revolution in response to our increasingly distracting environment of algorithmically driven media and AI dependency.

Cal Newport has influenced a lot of my thinking on cognitive fitness; he explores the idea in his “The Deep Life” podcast. In this linked episode, Cal outlines a smattering of protocols one might incorporate to their daily life to promote cognitive fitness. This inspired me to unify this into a single program that anyone can use to measure progress and results from their own cognitive fitness routine.

The program: 12 weeks of progressive overload

In weightlifting and bodybuilding, the idea of progressive overload is as old as the sports themselves. Try to add a little weight or an extra repetition every time. This method of incremental improvement is both measurable and guarantees results. I’ve had good results with this kind of program in the gym, so I wanted to see if it could translate to my work. The goal was simply to measure and increment the amount of time I spend in deep work each day. I started with sessions only 45 minutes in length, with only 10 sessions each week (only 7.5 hours of deep work/week), and worked my way up to 13 sessions of at least 75 minutes each (16.25 hours of deep work/week). This may not seem like much relative to a 40-hour work week, but I think you’ll find that isolating time for deep work is more challenging than you expect.

There are a few rules about what I count as deep work:

1. The time is uninterrupted - no notifications
2. Your phone is off or out of the room
3. You have set up the minimum required environment to engage the task at hand. No extra open tabs or programs
4. AI serves only to facilitate deep work. No context-hopping or mindless riffing.

This means you’ll need to leave your coworkers’ messages unread, and even worse you’ll have to leave your long-running coding agents waiting on your assistance. But reducing context switching and avoiding digital multitasking is shown to improve performance efficiency and overall cognitive health.

I also added some low-time-cost habits to my program to help with transitions and maintain the ability to focus throughout the day:

• Meditation (shown to improve focus, attention, and memory over time)

• Fill up water bottle (sounds silly, but I’m a chronically dehydrated individual)

• Micro-workout (pull-ups, or a brisk no-media walk)

• Put my phone away (I do any needed MFA first thing, then put it on a charger in another room).

I only track these once each day, but will often do multiple micro-workouts to lead into a context switch.

In order to track deep work sessions and my four low-cost habits, I developed this spreadsheet using Claude Opus to track progress over the course of 12 weeks. Please feel free to copy and use it as you’d like.

My personal results

Fortunately, cognitive adaptation can be achieved relatively quickly. I saw improvements in my ability to focus in just a couple weeks. Initially I was craving distraction fewer than 30 minutes into a task, but by week 3 I was easily recording sessions of 50+ minutes. By the 10th week, my average deep session duration had surged to around 100 minutes.

Cal Newport suggests that deep work capacity is finite, and that realistically it probably tops out at 2-4 hours per day, with the highest quality work done in 60-90 minute chunks. So, my next endeavor may be to increase the intensity or productivity of my deep work blocks, particularly around writing.

I did not include daily reading on my spreadsheet, but that is another pillar of training cognitive fitness, according to Cal Newport. Currently I’m reading The Book of The New Sun by Gene Wolfe, who also helped invent the machine that makes Pringles.