Cognitive Atrophy

To “atrophy” means to diminish or even waste away, and in the context of biology it generally refers to the reduction in size or effectiveness of some kind of tissue or cell.

If you break your leg and are unable to use it for several months, the muscles in that leg might atrophy from disuse. Likewise, if you don’t use a particular skill or collection of skills—let’s say those related to dancing—for years, there’s a chance that your muscles and tendons related to that activity will diminish, alongside (potentially) the structural integrity of your bones, and even the parts of your brain that provide the muscle memory for performing various moves.

The human body (and most biological systems) also tends to atrophy with age: our brains start shrinking around age 35 (at a rate of 0.2% per year) and our muscle tissue starts to diminish around that same time; around 3–8% of our total muscle mass lost per decade, from age 30.

While it’s possible to stave off some of the most deleterious aspects of this process, at least for a while, this sort of diminishment is a natural component of biological systems that lack built-in failsafes (reverting to a juvenile state and going through a lifetime of development all over again, for instance).

Unfortunately, many facets of modern life seem to amplify these atrophic impacts. Physical inactivity, obesity, chronic diseases (including things like cancer and autoimmune disease), and unhealthy diets can all speed-up this process, and can make its consequences more pronounced.

Interestingly, a recent study conducted by Microsoft found that using artificial intelligence systems (like OpenAI’s ChatGPT, but also Microsoft’s own Copilot AI tools) can lead to what’s called “Cognitive Atrophy,” with time.

This study suggests that shifting cognitive load—thinking and creative tasks that stress our brains, in the same way lifting heavy weights stress our muscles—can lead to cognitive atrophy.

Just as having a robot that lifts everything we might otherwise pick up throughout the day (including the weights at the gym) would probably lead to muscle atrophy, using these tools so that our brains don’t have to do as much lifting can diminish our thinking abilities in measurable ways.

If we rely on AI tools for most or all of our heavyweight cognitive tasks, we may, with time, lose our critical thinking skills in particular.

Our brains may become accustomed to shifting such burdens to these AI tools, and may then reorient their capacities in that direction, with the expectation that the difficult stuff will tend to be taken care of, externally (a process called “cognitive offloading”).

The implications of this finding are vast and varied, as on one hand, creating a workforce of people who don’t have to do cognitive heavy-lifting could be good for society (more people capable of doing big-time cognitive work), and it could be good for business in the sense that business owners no longer need to find and hire as many specialized, deep-thinkers (and they can probably pay employees less, as a consequence, too).

On the other hand, people who do knowledge work of any kind (which is a lot of people in the modern world) may find themselves unloading more and more of their responsibilities to AI tools (which feels good!), but in doing so they may be reducing their own cognitive capacity and decaying their skillset, which in turn could reduce their value to their employer, and perhaps even lead to them being replaced by lower-skilled, lower-paid workers, or automated AI tools (which feels less good).