so, about that study on chatbots and using our brains...

No, AI isn't making people dumber. It's just a symptom of a much larger problem.

According to a hyper-viral new study from MIT, using chatbots is quickly and surely making you a slack-jawed drooling dumb dumb who’ll soon have trouble tying their shoes without at least two concussions in the process. Or, at least that’s how some media outlets and a lot of staunchly anti-AI folks online are presenting the results of an otherwise very small, very nuanced study which adds to a growing body of work on how we handle automation to raise important questions.

Here’s the setup. A group of 54 people aged 18 to 39 were told to write essays with either the help of a search engine, their own brains, or a search engine while having their brain activity measured. Unsurprisingly, those who just had the chatbot spit out an essay had the worst recall and used their brains the least, a pattern they saw over four sessions spanning four months.

So far, so normal as far as a study goes. But after tabulating the data, this is when the whole thing took a strange turn. The lead investigator, Nataliya Kos’myna, decided to publish it directly to the arXiv pre-print repository and publicize her findings before a formal peer review process because she found them so startling and alarming, which is often a red flag in academia.

I say that because academics usually tend to do this if their paper either seems likely to be rejected, or makes serious leaps to conclusions they’re not sure will be given a pass by the reviewers. This is not exactly the case here as her stated motivation was to prevent lawmakers from dictating that schools use AI — or as our new Secretary of Education says, A One — in every classroom so they don’t have to hire more teachers and give them much needed resources. Which, to be fair, they are.

digitize, outsource, shortcut

But even if we downgrade this red flag to a yellow given the rhetoric around AI in class and the delusional belief of tech CEOs that chatbots and the web could replace most teachers, in spite of the massive worldwide drop in tests scores when we tried a really similar experiment during COVID, and take this study’s results as a given, there’s still a very serious question of what exactly we’re seeing in the data.

Yes, of course those who outsource writing an essay to a chatbot aren’t going to put a lot of cognitive effort into the task. This is the same thing we see when using mapping and directions apps, or using smartphones to manage phone numbers, or any gadget and app that figures something out for us immediately. Subjects who did the work had to pay attention and stay engaged had no other choice but to use their full brainpower while the ChatGPT users didn’t have to care.

And that’s the crux of the matter. What Kos’myna’s and similar research, along with an ocean of frustration from teachers and college professors while the same people who actively want to push us into a cyberpunk dystopia are actively funding a startup that help you cheat on everything from exams to job interviews with AI, shows us isn’t that using LLMs is making you dumb, it’s that there’s a societal deficit of caring about both the value of learning and knowledge, or following our rules.

We’ve taught generations of students that outcome is the only thing that matters, and that there are usually very few, if any, consequences for faking it ‘till you make it, lying, or behaving like a belligerent asshole. As long as you get yours, it’s all good. So, yeah, load that assignment prompt into ChatGPT, get a subscription to cheat on tests or job assessments with an LLM. You do you. It’s not like it matters anymore.

how to teach all the wrong lessons

Didn’t you hear? Normal day jobs are a scam for the poors, education is outdated and overrated, and some finance bro with a podcast and a pyramid scheme can teach you everything you’ll ever need to know, from money, to relationships, to fitness? If not, all you need to do is open any social media app and look at any right-leaning content for Gen Z and younger for five minutes. This message will dominate your feed by the end of the day from here on in.

You’re not going to fight that attitude by banning AI in classrooms. You’re not going to instill work ethic into kids who see that hard work gets them nowhere fast and that no one really cares about what they know, just how much money they can make, or what credential they can toss at a recruiter by telling them that AI makes them dumb.

They don’t care. Brains aren’t helping them. They may as well embrace the path of all those “dumbasses” with Lambos and mansions instead of the burnt out gifted kids in studio apartments in enough student debt to buy an investment property somewhere in the Caribbean to rent out as an AirBnB.

If you don’t want kids using AI and actually want to engage them in learning, you’re up against a society that does not value what you’re trying to do anymore. You will need to tear up Bush The Lesser’s No Child Left Behind which refocused education around testing rather than engagement, curiosity, and alternative pathways, the worst way to run an educational system according to experts. You’ll also need to show that being a scholar is rewarded rather than just punished by authoritarian regimes on a rampage, and can also yield a fulfilling, adventurous life.

That’s a lot more difficult than repeating the same mantras and platitudes about those no longer extant values of hard work, perseverance, and love of learning, or claiming a device that automates a task you don’t want to do for you makes you dumb. But very, very few of us seem ready and willing to have this conversation, or even know that we need to have it in the first place.