The Good, Bad and Ugly of ChatGPT for Your Brain and the Planet — The Tradeoff of Using AI to Write
Food for Thought from the Alliance for Sustainability
By Danbee You, Alliance Intern from Washington University, St. Louis ‘28
Admit it. We’ve all been confronted with a vapid work document, a mountain of unanswered emails, or a long essay assignment, and instead of sitting at our desks, rolling up our sleeves, and tackling the workload, we type “ChatGPT” into our browser.
The world is becoming increasingly reliant on ChatGPT and other AI tools. More and more Americans are opting for the polished, quick, perfect answers of ChatGPT over our own flawed writing that took triple the time and energy to develop.
As a college student, I admit I’ve used my fair share of AI to create study guides, summarize textbooks, and complete homework assignments, as have most of my peers. After my friend got caught using AI for a study abroad application, she used ChatGPT to write her apology email.
It is free, quick, easy, and accessible, with ostensibly minimal repercussions, right?
Of course, there’s the issue of stealing from human artists, the massive environmental impacts, the murky use and storage of personal information, the international race to dominate AI, and the ever-lingering fear of an AI takeover. But other than all that, the direct effect of AI on your life seems to be a net positive.
The Alliance has previously covered the challenges of AI and its data centers from a water and energy standpoint but has not addressed some of the other underlying concerns. This is the first serious critique documenting the significant challenges AI poses for brain functioning.
The Urgency of Sharing the Dangers of AI
A new MIT study published in June 2025 is one of the first major studies evaluating the direct effect of AI on our brains. They focused on Large Language Models (LLMs), AI systems that specialize in human language and writing.
The study is not peer-reviewed and has considerable limitations, such as a small and narrow sample size. However, Nataliya Kosmyna felt compelled to release the study before the full peer review due to the breakneck speed at which AI is being developed and adopted.
She shares, “I am afraid in 6-8 months, there will be some policymaker who decides, ‘let’s do GPT kindergarten.’ I think that would be absolutely bad and detrimental. Developing brains are at the highest risk.” As the idiom goes, you cannot put toothpaste back in the tube.
The Good and Bad of AI for Writing and Mental Functioning
The MIT researchers recorded the brain activity of three groups as they wrote essays with the assistance of one of three tools: The LLM group used ChatGPT, the Search Engine group used Google, and the Brain-only group used nothing.
The overarching conclusion? Over-reliance on AI when writing has detrimental, long-term effects on one’s cognition. The results highlight the cognitive sacrifice behind ChatGPT’s friendly “Ask anything” invitation.
Good News: AI is Easy and Helpful
The participants who used LLMs were less frustrated and less exhausted, meaning they could engage with a task more thoroughly, be more productive, and work for longer.
Short-term, AI is like the overachieving office intern who is eager to please and endlessly reliable. You could offload entire projects and make ChatGPT sift through pages of droning PDF files while you sit back and watch the sentences materialize before you. It is so easy, so accessible, and so not worth it.
Bad News: Two Major Challenges From AI in Memory and Learning
Long-term, the detriments begin to manifest in memory and learning. There have been times when I’ve submitted a paper using AI and struggled to recall the information I’ve just written about. New research supports that memory and learning suffer when writing with AI.
1. Writing Without Remembering: Memory Recollection
After relying on ChatGPT to write essays, LLM participants struggled greatly to recall a single line from their freshly written essays. About 88% of the Brain-only group and 83% of the Search engine group were able to provide an accurate quotation, while no one in the LLM group could.
Many profess the potential for AI to be used as a tool to facilitate learning, but these results suggest that the writing you receive from Chat bounces off your surface instead of being absorbed into deep learning. As you’ll see, it can even hinder future learning by weakening neural connectivity.
2. Brain “Conversations” Go Quiet: Neural Connectivity
As Nataliya Kosmyna, the leading author of the MIT study, describes it, neural connectivity is looking at “Who is talking in the brain and how much talking is happening.” These “conversations” build a foundation for future learning, meaning the struggle of conjuring original sentences makes the task easier and better in the long run.
Out of the three groups, the Brain-only participants demonstrated the highest neural connectivity and the strongest and most distributed networks. The Search Engine group showed similarly high results, indicating that old-school writing tools do not have nearly the same detriments on our brains as AI.
The LLM group results presented the most concerning findings: they exhibited the weakest connectivity, the lowest brain engagement, and “consistently underperformed at neural, linguistic, and behavioral levels.” While the Brain-only group had 79 neural connections, the LLM group had just 42.
Even when the LLM group switched to writing without AI, their neural connectivity was consistently weaker than the Brain-only group.
The Psychological Costs Add Up
These two deeply concerning aspects of AI lead to an overall detriment to brain functioning called cognitive debt, in which the brain does not have the same capacity as it once did. AI use leads to a lack of brain engagement, resulting in issues of memory recollection and neural connectivity.
This results in a troubling trend: accumulation of cognitive debt.
The simple analogy is like how an old car needs to be taken on a drive every now and then to keep the moving parts working. We must exercise our brains to keep them sharp. Neglect it for long enough and the damage piles up. That’s the effect researchers are finding from extensive AI use for writing. Thanks, AI.
Over time, your brain may forfeit the skills required for independent thought. Critical thinking and creativity suffer, and users become more vulnerable to manipulation. By letting down these guards, users become susceptible to internalizing the superficial and biased perspectives that ChatGPT offers.
Microsoft corroborates this finding in their own study: “higher confidence in GenAI is associated with less critical thinking, while higher self-confidence is associated with more critical thinking.” How ironic and scary it is that one of the world’s largest promoters of AI is saying that more AI will bring less critical thinking.
“But ChatGPT is a Better Writer Than Me.”
Does this quote resonate with you? Most of us lack confidence in our writing compared to ChatGPT. I’ve lost count of the number of times Chat has perfectly worded a thought I couldn’t articulate. It left me feeling deflated and unconfident. Over time, it seemed easier, even more responsible, to let Chat take over on something I could never match up to.
The finding from the Microsoft study documents how AI undercuts our self-confidence and interferes with the belief in our own writing ability. You need to believe in your writing. If you truly believe your writing is irredeemably awful, still trust in the value of your perspective.
When English teachers assessed the writing of the LLM group, they acknowledged that AI-written essays had nearly perfect use of language and structure. They sounded academic and developed a topic more in-depth than others.
However, the teachers also found the ChatGPT-written essays to be frustratingly “soulless” and “empty,” lacking nuance, personal insight, or individuality.
The teachers consistently valued these traits — nuance, insight, individuality — more than perfection, and who wouldn’t? Have you ever tossed away a birthday card because someone put a comma after “Dear”? Would you ever prefer the air-brushed look of AI art over a friend’s eraser-streaked drawing?
Yes, ChatGPT may be a better writer, technically speaking, but we don’t need more grammatically perfect sentences. We need perspective and soul.
The writing ChatGPT offers you is also distributed to millions of users every day. As we integrate AI writing into our society, we make room for soulless, albeit perfect, writing over imperfect and weird voices. The diversity of writing styles is what makes each special and distinctive. Conforming to a homogeneous style of writing means the loss of culture.
The Burden of AI on Our Planet
In addition to damaging our cognition and critical thinking, each query consumes enormous amounts of energy and water that strain our environment and resources.
By next year, data centers are projected to use 1,000 terawatts of electricity, roughly equal to Japan’s total consumption. LLM queries in particular require 10 times more energy than non-LLM prompts.
Moreover, data centers from Google, Microsoft, and Meta withdrew 2.2 billion cubic meters of water in 2022, about twice the annual water usage of Denmark. This massive water footprint forces people to compete with AI for limited water resources. For example, Google’s data centers were using a quarter of all available water in The Dalles, Oregon.
Are There Positive Ways AI Can Be Used to Write?
While these studies underscore the serious threats posed by AI writing, it’s clear that AI is unlikely to be wholly removed from our society. It underscores that we use AI in the right way.
While the group that went from LLM to Brain-only writing had persistently weaker neural connectivity, those who went from Brain-only to LLM writing exhibited stronger brain connectivity.
Purposeful application of AI could even aid environmental efforts by improving climate models, finding efficient ways to cut energy consumption, and sorting through air pollution data. AI-run smart homes could even reduce household CO2 consumption by up to 40%.
This suggests that AI can bolster our work and improve productivity if used correctly. Specifically, those with strong baselines for critical thinking and creativity will flourish with the aid of AI. However, just like any muscle, if we don’t exercise our critical thinking skills by writing without AI, cognitive atrophy is the result.
Where is the Human in AI?
As the world becomes automated and streamlined, as Ring doorbells replace neighbors’ faces and as the robotic self-checkout station voices talk over cashiers, creative expression was supposed to be an untouchable haven for raw emotion and original thought.
In every sci-fi movie in which humanity battles against robots or algorithms, the saving grace of the former is always emotion and imperfection. So will we forfeit this fight without a struggle? Will we surrender the privilege of language, the ability to communicate anything in the world, just because it’s hard?
How many times have you sat in the theater after watching a highly anticipated movie remake and thought, “The book was better”? The time, energy, and patience required to deeply consume something, to scan every letter of a black-and-white book instead of staring at a mega-screen of flying colors, makes the content more enriching.
AI-assisted writing is enticingly easy and literally a fingertip away. It feels good, like a hack, and no one has to know. But at the end of the day, no matter how AI may recycle words, it is simply a messenger using an algorithm to mimic us. We lose a bit of humanity and a bit of ourselves each time we pick the easy route and communicate through AI.
The beautiful one-liners that AI may throw into a message are a product of human insight, human emotion, and human struggle. You are not a fan of AI — you are a fan of humanity.