I use artificial intelligence every day. Apparently, this may be making me stupid.
Well, that’s as good an explanation as any.
But there are a few things to remember before concluding that we’re all reverting to caveman times.
First off, people are lazy. Evolution backs that up.
Or, to put it another way, evolution rewards efficiency. Why waste energy if you don’t have to? We started using tools to save wear and tear on our hands, backs, and bodies. We sharpened stones instead of tearing meat apart with our teeth. We invented the wheel because dragging things is a lousy way to spend an afternoon. We built machines because human muscles get tired.
In that sense, almost every technological advance is an attempt to conserve effort.
AI chatbots and agents are just the latest extension of that instinct. The difference is that now we’re conserving mental effort instead of physical effort.
You can argue that tools have made us physically weaker. But are we really going back to cutting lumber with stone axes? Harvesting wheat by hand? Building houses without power tools because it would be better exercise?
Of course not.
The question is what happens when the effort being saved is thinking.
We’ve spent thousands of years inventing machines to save our backs. We’ve now invented one that saves us from having to use our brains. That requires a little more caution.
Calculators didn’t make arithmetic useless. GPS didn’t make geography pointless. Spellcheck didn’t make knowing how to spell worthless. But all of them allowed us to stop exercising abilities we once had to use every day.
AI may simply be the most powerful cognitive-offloading machine we’ve ever invented.
The answer isn’t to throw it away. It’s to decide which parts of your brain you aren’t willing to outsource.
A recent MIT study prompted a lot of headlines and hand-wringing about AI making us dumber.
Researchers had 54 people write essays either with an LLM, with search engines, or without any outside tools. EEG measurements showed the strongest and most widely distributed brain connectivity among the people working without tools. It found moderate engagement among search users. But the weakest connectivity was among writers using an LLM. The AI users also reported less ownership of their essays and had more trouble quoting what they themselves had supposedly just written.
But the study comes with a pretty big caveat. It was small, and only 18 people took part in its crossover fourth session. It also looked at one very specific activity: essay writing. That’s not enough to establish that using ChatGPT causes permanent cognitive decline.
And the picture gets more complicated.
The same MIT group published another study this month involving 50 people learning nuclear-safety material. Researchers compared unrestricted chatbot use, a Socratic chatbot that offered hints instead of answers, and an adaptive tutoring system.
The unrestricted chatbot users actually scored better on the immediate test. But the researchers said that better short-term performance wasn’t necessarily evidence of deeper learning.
There’s also an interesting Wall Street Journal article about some pretty smart businesspeople discovering that they’re starting to talk like chatbots.
One executive caught herself opening a work meeting with what’s often called contrast framing, a linguistic device ChatGPT, Claude, and other AIs seem to love:
“This is not a morning meeting, it’s a reset.”
Other executives told the Journal that things they say in meetings are beginning to sound like the prompts they use with AI.
Look, language evolves. We don’t talk today the way we did 100 years ago. An English speaker dropped into medieval England might feel like a foreigner who barely knows the language.
But this may be the first language evolution driven by a language tool we created ourselves.
And that raises another question. If AI starts changing not only how we think, but how we sound, who’s training whom?
So if you’re worried about getting dumber because of AI, but you want to — or need to — keep using the tools, I think there are some practical guidelines worth following. That way you can keep telling people, “Stand by for greatness, for I am a genius!”
Here they are:
Never let AI become your voice. The machine is supposed to learn how I communicate. I’m not supposed to learn how the machine communicates.
Think before you prompt. Have some idea of what you believe or want before asking AI what it thinks.
Use it to challenge your thinking, not replace it. Ask, “What am I missing?” “What’s the strongest argument against this?” “Where does my reasoning break?”
Don’t outsource the blank page all the time. There is cognitive value in struggling to figure out what you want to say. Make your brain work.
Make AI explain things instead of just answering them. “Walk me through why” is more valuable than “give me the answer.”
Interrogate its answers. Ask for evidence. Check sources. Disagree with it.
Rewrite it in your own head. If you can’t explain what the AI just told you without looking back at the screen, you probably haven’t learned it.
Keep doing some things the hard way. Read whole books. Write paragraphs. Remember things. Work through problems. Mental exercise isn’t wasted motion.
That’s not just good advice. It’s a daily mantra.
Oh crap, that sounds like AI.
If this made you think — with or without artificial assistance — share it with someone who might enjoy it. And if you’d like more Archer’s Line in your inbox, subscribe at therobarcher.com. Paid subscriptions help keep this thing going, and they’re deeply appreciated.
AI has become my PCP. Btw, what is LLM?