Artificial intelligence (AI) is intrinsically linked to and associated with notions of innovation and a tech-driven future. Although it’s actually something we’ve already been using for a while, albeit in less obvious ways, it’s still something that ignites a great deal of excitement, especially as new gadgets and types of tech hit the market.
But, the other thing that’s been a consistent component of AI and the industry more generally is a certain degreee of concern and fear surrounding its capabilities and the ways in which it may potentially be used for nefarious purposes. In the most extreme sense, people worry about AGI and ASI (that is, artifiical generative intelligence and artificial superficial intelligence) and its ability to become as intelligent as humans (if not more, in the case of the latter) – thus, the potential for “robots” to get involved in some “shady business”, so to speak.
Or, potentially more realistically, especially with regard to the actual ability of technology at this point, the way in which humans may manipulate advanced AI technology to do bad, potentially criminal and unethical things.
However, according to a recent study conducted by MIT, those aren’t the only things we need to worry about these days. In fact, of significantly more concern is what researchers are calling an “accumulation of cognitive debt” as a result of consistent use of AI in a variety of applications – the main focus in this particular study, however, was AI-powered writing assistants that are based on LLMs.
What Does the MIT Study Tell Us?
The term “cognitive debt” is a little verbose – there’s no getting away from that – but essentially, what it means is that our cognitive ability is deteriorating. Consistently “outsourcing mental effort” – that is, asking AI to think and write for us – is making us less capable of doing that ourselves.
Basically, by making AI smarter and more capable, we’re removing our own need to perform these tasks that require thinking, and, as a result, we’re actually losing our ability to think as well as we used to. Of course, this is a pretty simplistic, crude way of putting it, but you get the picture – AI is making us stupid. Well, not quite, but that’s the gist of the sentiment.
So, what did the study actually do and test in order to come to this terrifying conclusion?
More from Artificial Intelligence
The MIT Study On Accumulative Cognitive Debt
This study, entitled “Your Brain on ChatGPT: Accumulation of Cognitive Debt when Using an AI Assistant for Essay Writing Task”, was all about looking into how AI tools may affect brain activity, learning behaviour and the quality of essays produced in the process as opposed to writing unaided or only with the help of traditional (more basic) search engines.
Participants were split into three groups: one group used AI, one used old-school search engines and one used only their own thinking (that means no internet or anything at all). They all wrote essays over three sessions, and eventually, in a fourth session, some switched groups.
During these sessions, the researchers measured the participants’ brain activity using EEG (electroencephalogram, a test that evaluates electrical activity in your brain), and the essays were analysed by teachers, AI (ironically) and language software.
The results showed that the more that people relied on tools (especially AI), the less active and connected their brains were during writing. Those who didn’t use tools had stronger brain engagement and felt more ownership over their work. AI users had the weakest brain activity and often couldn’t remember or explain their own writing.
Switching from AI to thinking-only caused a drop in brain activity, suggesting a kind of “cognitive debt” from relying too much on AI. In contrast, people switching to AI showed more memory use but still less than those who used only their brains. Thus, the conclusion that these researchers came to was that while AI tools are convenient, they may reduce learning, memory and critical thinking over time. The study warns that heavy reliance on AI in education could have long-term negative effects on how we think and learn.
Is AI Making Us Less Intelligent?
As AI tools are becoming more common in education and everyday writing, there’s growing concern about their impact on our thinking skills – especially with regard to children who, with young, impressionable minds, may be among the most effected in the long run.
The ease and speed they offer may actually discourage deep thinking, critical analysis and the effort needed to fully understand complex ideas. Instead of struggling through a problem or forming original arguments, many users now lean on AI to generate answers instantly. It can kind of be compared to how many parents and math teachers used to argue that you “shouldn’t use a calculator” because first, you won’t always have one with you (well, that didn’t quite go as planned) and second, if you always rely on a calculator, you’ll never be able to do the math yourself.
The same goes for AI and general cognitive skills – if we allow AI to do all the thinking, all the heavy lifting, for us, it follows that not only will we probably not want to do any hard work going forward, but we may actually not be able to.
Ultimately, it can lead to surface-level learning, where information is absorbed passively rather than actively processed. Over time, such habits may weaken memory, reduce creativity and limit the ability to construct clear, independent thoughts.
Where To From Here?
In the wise words of young Ron Weasley, can we panic now?
Well, I think “panic” may be a little dramatic, at least for now, but it’s definitely something we can’t ignore. While AI is a powerful aid, relying on it too heavily may shift us from being active learners to passive consumers of information.
The real challenge lies in finding a balance – using AI to support our thinking rather than replace it completely – so that our intelligence develops alongside technology, rather than being dulled by it.
The most concerning and time-sensitive issue to consider, however, is the effect that AI may already be having on children, because they’re the ones who are experiencing a broad rollout of a variety of AI-powered technologies that are aiming to make their lives easier. And, if this is how we educate children from the outset, the concern is that they’ll not only lose their ability to do and think of things themselves, but that, in the case of young children, they’ll simply never learn how to do them at all.
To avoid a gradual slide into a world of “Idiocracy” (the movie, but take that as you will), we need to pay close attention to what AI technology is being used by our children at school and in general life to make sure that they’re experiencing at least a balance of natural, old-school learning as well as a bit of the technically advanced, AI-type stuff.
This is yet another reason why it’s so important that AI regulation is current, keeps up with innovation and is implemented effectively.
