{"id":7148,"date":"2026-03-12T11:06:21","date_gmt":"2026-03-12T15:06:21","guid":{"rendered":"https:\/\/beomniscient.com\/?p=7148"},"modified":"2026-03-18T10:01:31","modified_gmt":"2026-03-18T14:01:31","slug":"the-atrophy-paradox","status":"publish","type":"post","link":"https:\/\/beomniscient.com\/blog\/the-atrophy-paradox\/","title":{"rendered":"The Atrophy Paradox"},"content":{"rendered":"\n

In 2013, a federal task force wrote a 267-page report that concluded<\/a> the following: pilots relied too often on automation and should be required to improve their manual flying skills.<\/p>\n\n\n\n

This followed a 2011 study showing many pilots<\/a> had trouble manually flying a plane or handling automated controls. <\/p>\n\n\n\n

A few years ago, the FAA released a much stronger statement on the topic<\/a>:<\/p>\n\n\n\n

\n

\u201cThe FAA urged pilots in coming years to sometimes hand fly \u2018entire departure and arrival\u2019 routes, or \u2018potentially the entire flight.\u2019 Such practices are supported by years of FAA research and Congressional directives. However, they generally are shunned by airline management focused on utilizing automation to improve fuel economy and maximize smooth rides for passengers.\u201d<\/em><\/p>\n<\/blockquote>\n\n\n\n

This is, by and large, a debate around automation and its effect on human performance and cognition. Though much more is at stake with air travel, we are dealing with the same thing in knowledge work right now. <\/p>\n\n\n\n

Generative AI, when first launched, became a way to \u201cdefeat the blank page\u201d or spruce up some copywriting.<\/p>\n\n\n\n

It then became a sparring partner.<\/p>\n\n\n\n

Of course, it completely restructured the craft of programming.<\/p>\n\n\n\n

And every day, it is getting both more powerful and more accessible (for example, it is functionally very easy to get Claude Cowork to do remarkable things for you even without \u201cprompt engineering\u201d or any advanced technical skills.). <\/p>\n\n\n\n

All of this has drastically improved the productivity of most knowledge workers, including your author. <\/p>\n\n\n\n

However, the ability to use AI tools well \u2013 let\u2019s say, for marketing purposes, in a way that produces better results than your competitors \u2013 is contingent on the very abilities that AI tools abstract away (writing, domain expertise, and at the very center of it all, critical thinking). <\/p>\n\n\n\n

In other words, a person who is good at writing and critical thinking and has domain expertise will get better results from AI than someone without those things, yet over-reliance on AI tools tend to degrade those very things. Thus the atrophy paradox<\/a>. <\/p>\n\n\n\n

The FAA understood something that most industries haven\u2019t caught up with yet: proficiency is use-dependent. Stop using a skill, and the skill quietly degrades, even if you don\u2019t notice it degrading, even if your outcomes stay the same, because the tool keeps compensating for the loss.<\/p>\n\n\n\n

Marketing has no equivalent mandate. And we\u2019re adopting AI tools faster than aviation ever adopted autopilot.<\/p>\n\n\n\n

The paradox isn\u2019t that AI tools are bad. They\u2019re not. They\u2019re extraordinarily useful. The paradox is that the tool\u2019s effectiveness depends on a skill that the tool itself erodes. The better the autopilot, the worse the pilot. Unless you deliberately maintain the underlying capability.<\/p>\n\n\n\n

The Brain on Autopilot<\/strong><\/h2>\n\n\n\n

There\u2019s a growing body of research on what happens cognitively when we hand work over to machines.<\/p>\n\n\n\n

In 2025, researchers at the MIT Media Lab used EEG<\/a> to measure brain activity while participants wrote essays, some with AI assistance and some without. <\/p>\n\n\n\n

LLM users showed less brain connectivity, weaker linguistic markers of deep processing, and lower overall cognitive engagement. <\/p>\n\n\n\n

A separate study published in Societies surveyed 666 participants<\/a> and found a significant negative correlation between frequent AI tool usage and critical thinking abilities. This was basically due to cognitive offloading (the act of transferring mental work to an external system). The more you offload, the less the underlying cognitive muscles are engaged.<\/p>\n\n\n\n

There\u2019s a principle in learning science called the generation effect<\/a>: you remember information better when you generate it yourself than when you passively receive it. Writing activates more brain regions than reading. Feynman Technique<\/a> in a nutshell. Struggling with a problem encodes the solution more deeply than being handed the answer. <\/p>\n\n\n\n

A very interesting 2000 study by Eleanor Maguire at University College London<\/a> showed that London taxi drivers (who must memorize 25,000 streets to earn their license) had measurably larger posterior hippocampi than the general population as well as bus drivers (who drive the same streets but follow fixed routes). <\/p>\n\n\n\n

And when cabbies retired and stopped navigating, the hippocampus shrank back (like a muscle you don\u2019t use).  <\/p>\n\n\n\n

Now, when I use GPS, I believe that I am at a distinct advantage to not using GPS. I remember (vaguely) printing out maps from Mapquest in the earliest days of my driving career.<\/p>\n\n\n\n

So it may very well be the case that you don\u2019t need to build a manual muscle for everything (and in fact, it would be very beneficial to abstract away certain cognitive processes to focus on more valuable and generative things). <\/p>\n\n\n\n

The science shows a pretty clear pattern, though, and it poses some interesting questions for SEOs about intuition, skill, and domain expertise:<\/p>\n\n\n\n