So, is AI making us dumber or not? The truth is, it's still not entirely clear
For quite some time now, the internet has been flooded with the belief that prolonged use of artificial intelligence systems such as ChatGPT could, to a greater or lesser extent, reduce people’s ability to think and solve problems independently. That is precisely why Wired recently addressed the issue, revisiting a study conducted by researchers from some of the most prestigious English-speaking universities.
While it is true that the impact of artificial intelligence on everyday life and on people’s cognitive abilities has sparked countless debates, both within and outside the tech industry, the reality is that — despite growing interest in the field — the available research remains fairly limited. At this stage, it is still difficult to predict how AI might genuinely affect human thought, reasoning, and decision-making processes over the long term — assuming it does at all.
Is it too early to understand AI’s effects on people?
@david_epstein Brain first. AI second. An MIT study found that students who started essays with ChatGPT remembered less (almost nothing, really) and showed weaker reasoning. When they thought first and used AI second, though, their writing and recall improved. Use the tool to amplify your thinking, not replace it.
original sound - david_epstein
Most experts working in the field of artificial intelligence argue that these studies, while certainly interesting, should not be interpreted as definitive proof of this technology’s impact on human cognitive abilities. Many of the studies published so far are based on small sample sizes, experimental settings, and tests that hardly reflect the way people actually use chatbots in their daily lives. To fully understand AI’s long-term effects, academics themselves believe that broader, more diverse studies conducted under more realistic conditions will be necessary.
Many of the artificial intelligence studies amplified by the media and spread across social networks — such as the one presented in June 2025 by the Massachusetts Institute of Technology (MIT) in the United States — are also “preprints”, meaning preliminary studies that have not yet completed the peer-review and validation process required by the scientific community. In academia, findings are normally reviewed by other researchers before they can be considered reliable — and this verification process can take a considerable amount of time.
The limits of AI experiments
Most studies examining the potential impact of artificial intelligence on cognitive processes — including the one covered by Wired — tend to focus on immediate or short-term effects, without providing meaningful insight into the long-term consequences for memory, learning, or reasoning skills. Furthermore, the very structure and methodology of many AI-related experiments make it difficult to observe how the continued use of tools such as ChatGPT could gradually influence the way people process information, make decisions, or develop new cognitive skills over time.
Since artificial intelligence systems became widely adopted on a mass scale, largely thanks to the success of OpenAI’s products, millions of people have started interacting with chatbots in far more immediate and natural ways. In most cases, however, these tools are used to solve relatively simple tasks — such as summarising texts, among other things. For this reason as well, it remains difficult to determine the real impact they may have on human thinking, both from a neurological and psychological perspective. After all, understanding the long-term cognitive effects of technology on people has always been an extremely complex task, even in the case of far more established tools such as books, television, or even the Internet itself — technologies whose possible mechanisms of influence on the human brain are still not entirely understood.
