Use It or Lose It

The first time I found out about the use-it-or-lose-it properties of the human brain came when I was discussing my upcoming LASIK appointment with my surgeon.

Before LASIK, I was nearsighted in both eyes, but my left eye was a lot worse than my right eye. One of the last pairs of glasses I ever had to buy had different types of lenses for each eye: the special lightweight (expensive) lens for the left, and the regular el cheapo version for the right to balance out the weight on both sides.

The surgeon told me he couldn’t get my left eye to 20/20 vision. Not because his technique was bad, but because my brain had simply shut down that possibility. It required too much energy so my brain was not having it. And after the surgery, I could see, which was amazing!

Another example: I used to be the navigator in my family. My dad drove, and I was in charge of the map. (My mom was terrible with directions.) He looked at the route ahead of time so he knew generally where he was going. I kept an eye out for roads or towns that we passed to see where we were on the map and how close we were to the destination.

When I first moved to LA, in 2004, I bought a Thomas Guide to get around. Californians know what I’m talking about! It was a map book that detailed Los Angeles, and once you understood how to use it, it really helped you navigate.

Well, of course now we have our phones. I prefer Google maps to the GPS in my car, which is very slow and not intuitive. The other day I thought I’d try to use just the map (not the directions) to get to where I was going, and… I failed. 

Because I’ve been outsourcing navigation to Maps, I can no longer find my way without help. I stopped using that particular skill, and I’ve now lost it. 

I bet you can guess where I’m going with this!

Why the brain shuts things off when you don’t use them

Our bodies, like all animal bodies, try to preserve energy when they can. Energy requires fuel, and we humans (most importantly, our brains) developed in a time and place where food was scarce. 

There were no farms, much less grocery stores, and we ate what we gathered and killed. That’s why our bodies store fat so easily: to have energy available when we need it during a time of scarcity, though we don’t have scarcity any more.

Not only is energy in general “expensive” for our bodies, our big brains are extremely expensive. It’s responsible for about 2% of the body’s weight, but 20% of its energy expenditure. 

The human brain (like all animal brains) is a conservation machine trying to be efficient with its energy. That’s why if you don’t use a function, your brain will shut it off: too much energy for something you’re not using. 

Another way to think about it is at the neural pathway level. You’ve probably heard the phrase "neurons that fire together wire together”, especially when discussing the formation of habits. Basically, when your brain keeps using a certain pathway, that specific neuron arrangement grows stronger because the neurons are firing, and thus wiring, together.

What happens to pathways your brain doesn’t use? At night, during sleep, your brain strengthens the pathways you’ve been using and prunes back the ones you haven’t, in the name of energy conservation. So, when you stop using a function (the ability to navigate, or see with a certain precision), the brain prunes those pathways back.

I could bring back the navigation if I started using it again, because my brain would start strengthening those pathways. Whether or not I want to start navigating like I had to in my 20s and 30s is a different question. But there’s no real way for me to get to 20/20 vision since I can’t direct the process consciously. 

Given the way the brain works, there are probably a lot of functions you don’t want to outsource to AI

I know a lot of people don’t understand how AI works, and the only point I’m going to stress here is that the LLMs you use (Claude, ChatGPT, Gemini, whatever) get things wrong and make things up. These errors are often called hallucinations, but since AI is not conscious, a better word is “confabulations”.

It does not matter which version of the LLM you’re using, or how many years of “training” have gone into the models. The AI companies themselves have told us that these hallucinations and errors are inherent in the process.

That means you will always have to check the output from AI. 

Always. 

Every time.

No matter how you use the LLM, I cannot stress enough that you cannot let the results loose in the world without checking them.

But if you outsource your thinking to AI, how will you be able to check its results? Especially since it is designed to blow smoke up your ass about your amazingly intelligent questions? If you don’t have a good handle on what the output should look like, because you no longer have the ability to think critically about what the output is, how do you know the results are good?

And maybe you don’t want to check the output, because you’re the business owner and you have better things to do with your time. That’s a fair argument. Maybe you have a junior person who’s going to be responsible. They might have used AI for their papers in college, and lack critical thinking ability. 

How are they going to check the output and make sure there are no errors? They can probably do it, but because they don’t have the background, it will probably take them a lot longer than it should.

As you might know, I was a financial planner for years. By my last job, I could look at a client’s expenses, their income, how much they saved every year and how much they had already saved and know without running the financial software program whether or not they could make it safely to retirement. 

So if someone had AI doing the job, I could tell relatively quickly if it had made a major error. I wouldn’t have to dig into the numbers very far, because I could tell just by looking at the input.

How did I get there? By putting in the reps. I had built so many plans that I knew what factors tended to derail plans and what made them robust, before I even entered a client’s info into the software.

No one who only uses AI will be able to do that. You can get a result. But if you don’t know how to produce it without the help of AI, you don’t have a clue whether it’s a reasonable result or not.

Using AI might help in the short term but derail your “golden years”

Studies are already showing that people who use ChatGPT to write things can’t remember what they wrote or the arguments they made. They also think they know a lot more about certain subjects than they actually do, which may be a result of the sycophancy mode these models have turned on. (Dunning and Kruger would like a word.)

If you ask longevity experts for their best practices, in addition to advice on managing stress/exercise/nutritious food, they tell you to keep your brain active. Learning new dance steps. Having friends you get together with (no, texting and replying to them on “social” media doesn’t count.) Doing puzzles, etc.

I haven’t seen anyone talking about this. Although we do have data on how AI affects the working population and children, we don’t really have the info for retirees.

So I think there is at least the possibility that if you use LLMs to solve problems, think for you, and write for you, that you could very well end up in retirement unable to think. Which may or may not have an effect on your likelihood of neurodegenerative diseases like Parkinson’s or Alzheimer’s.

But even if I’m wrong about that, you will absolutely be more susceptible to grifters and con artists. Many of the elderly are more susceptible anyway, but as a result of offloading your ability to think and make good decisions, you won’t have the critical thinking abilities that prevent you from succumbing. So you could very well run out of money because you’re not able to see the con man coming and you get swindled out of everything you made.

Again, not saying this will happen to everyone. Do you want to run the risk? I don’t!

I’m not saying that you shouldn’t use AI at all. I think you really need to use your discernment to decide what you’ll use it for and what you won’t. Remember that when you’re considering how much it speeds things up, you have to factor in the time it takes to check for errors.

Because they’ve been “trained” on material that’s available, LLMs are good places to go when you need the wisdom of crowds.I used an LLM to analyze the 1-star and 5-star rated books in my category when I was writing my book on getting $hit done, to find out what people liked and disliked about books in that category, for example.

Recap (tl;dr):

Letting AI do your thinking for you will likely result in your losing the ability to think critically about your business and what’s going on in your world. There are ways to use AI that can be helpful without degrading your cognitive ability, but you have to discern where it will be helpful.

Feel like you have way too many tasks but you don’t want to dump everything on AI? (A wise move, I must say! There’s my sycophancy for the month.) Anyway, I can help you get on top of your business without spending a lot of time or money and without necessarily adding more software apps. Click here to schedule a time to chat with me.

Illustration by Alghozy on Unsplash.