Dana O'Driscoll If we look across nature, the species that are resilient, cunning, and adaptable are often the most successful. Cockroaches, rats, pigeons, and raccoons may top the list of extremely adaptable species–they have adapted even to urban environments and have found ways of thriving despite adversity. But the other species that I’d put on this list is humanity. Humans, by nature, are curious, we learn quickly, and we are resilient. But today, humans are becoming less resilient, less adaptable, and ever-more dependent on our technology. I’ll share this with a simple example. I have a friend who is extremely resilient–if something is broken, she will figure out how to fix it, even if it takes her a while to do so. If she needs something, she is always able to come up with creative solutions using the limited resources she has at hand. She is constantly learning and growing, and because of this, she has a very broad skillset due to all of these examples. If I am struggling with something, she’s the person I call because she will always have a solution. But I also have another friend who is literally the opposite. She has cultivated in herself a sense of learned helplessness. She can’t be bothered to do anything, and she always manages to get someone to do everything for her, to pay people to do it, etc. She loves AI and has AI do everything for her–AI is her therapist, AI creates her grocery list and menu, and AI even controls parts of her home. The problem is, she’s stagnating as a person. She doesn’t allow herself to grow through challenges, and she’s one of the least resilient people that I know. These simple stories outline a core point I want to share today–that we lose something important, something that fundamentally makes us human, as we relegate more and more of our cognition, creativity, and struggle to the AI Machine.
In my last post, we explored the explosion of AI Data Centers and why AI is terribly bad for the planet. I was saddened to hear from so many on how they are also fighting data centers, fighting the use of AI at work, being forced into massive AI use, and trying to stay human and connected in these times. Friends, I hear you, I see you, and I stand with you hand in hand against the insanity. In today’s post, I’m going to be putting on not only hat I always use when I write this blog–that of a druid, but I’m going to put on another hat you don’t see as often–that of my expertise as a learning researcher who studies long-term development of people. And with those two hats, I’m going to make the case that not only is AI bad for planet, it is also very, very bad for people.
Today I’ll again be talking about GenAI–this is AI that is widely available to consumers and is now widely adopted by companies: this is AI that is designed to generate output: writing, business communication, reports, data analysis, art, video, music, and complete many tasks originally relegated to humans. The most common of these systems are things like ChatGPT and Claude, but so many specialized systems are also making their way into daily use. The goal of these systems is to allow a human consumer to prompt the AI to do something, and then the AI will produce copious amounts of output. Increasingly, AI companies are now creating AI assistants that read all your files and have access to your entire work profile and do a lot of tasks for you. Of course, what these corporate AI systems are really doing is learning your job well enough to do it without you. Just like many other modern digital technologies (social media, games, etc), GenAI systems are also designed to be all-encompassing, addictive, and keep you consuming more.
GenAI Is Bad for People and Rots Your Brain
GenAI is bad for our own humanity: our skills, our expertise, our happiness, and our mental health. I’m going to share a number of new studies as well as a discussion that comes from my own research on this area. As a bit of background here: I have been a learning researcher for over 15 years professionally: my research focuses on understanding how people develop expertise and learn over long spans of time, and how writing is part of that equation, and how we can build better writing support. I’ve published over 70 peer-reviewed articles on the topic and have a forthcoming professional peer-reviewed book on the topic (and you can see my Google Scholar page). A lot of what I’m sharing here comes an article that I wrote about my concerns surrounding generative AI and expertise development for educators–and that’s currently undergoing blind peer review.
Much of AI use includes what is called “cognitive offloading,” where individuals choose to have GenAI complete both mundane and challenging tasks traditionally done with human brainpower (Shum, 2024). You can think about it this way: you have a short article you have to write for a local newspaper, and you haven’t done that before. Pre-AI, you would have to learn yourself how to write it–you’d look at models, you’d write a draft, and you’d get feedback. But now, in the age of AI, AI can do this for you. Rather than write it, you type a few sentences into AI and with a few more prompts, AI writes the bulk of it. Then you do a little more tweaking, and then give it back to AI to copyedit. Bam! Now you have your article, and it’s taken you all of 30 minutes, when it would likely have taken you several hours. No big deal, right? Perhaps not. Or, perhaps taking a few hours to write the article would have taught you a lot of new things, and you would have grown as a person. Now let me be clear–cognitive offloading one time isn’t a big deal–the issue happens when the effects become cumulative. Consider what would happen if you keep cognitively offloading, day in and day out. The one article you gave to AI to write was so easy to deal with, but now you have decided to use it for many other tasks. Those cognitive declines are going to be permanent. Eventually, you won’t be able to think your way out of a wet paper bag. Your brain requires use and training, just like every other part of your body. And in fact, this is being borne out by emerging research on the detrimental effects of AI.
Emerging research is demonstrating that, unsurprisingly, cognitive offloading to GenAI has serious consequences for critical thinking skills. Using mixed methods surveys and interviews of 666 participants, Gerlich (2025) demonstrated significant correlations between the frequency of AI use and the decline of critical thinking. What this means is that the more people use AI, the less critical thinking they demonstrate. Further, the younger the participant was, the more the decline of cognitive function occurred with higher AI use. Also this year, researchers from MIT (Kosmyna et. al., 2025) explored essay writing under three conditions: brain-only writing, LLM-assisted writing, and writing with the assistance of a search engine–they observed people writing with these conditions over four months. Using EEG brain scans, they found that the essays of the three groups were comparable in quality. However, people’s brains weren’t. Cognitive activity significantly decreased with the use of AI tools. Those using AI LLMs demonstrated the lowest cognitive activity across the brain, and over time, the LLM users performed the lowest on behavioral, neural, and linguistic tasks (translation = AI rots your brain!). And yet, we have missives from all sorts of educational institutions that we should be teaching and using more and more AI (such as a recent one from the Department of Education in the USA). They do this without knowing or caring about the implications for human learning.
Here’s another metaphor: let’s say you really like cheeseburgers from your local fast food chain. If you go eat a cheeseburger once a month, the cumulative effects of that are pretty minor. But let’s say you go and eat a cheeseburger and fries every day (just like the guy from the movie Super Size Me did). Pretty soon, that’s going to catch up with your physical health. I’d argue that, based on all of the available evidence, using AI on a daily basis and cognitively offloading more and more tasks to AI is pretty much like stuffing your face full of cheeseburgers every day. Rather than affecting things like your cholesterol and weight, AI affects your cognition, happiness, skill, and focus.
Emerging research from the field of psychology suggests that even those people who are already experts in professional fields can easily grow dependent on AI and see detrimental effects. Naseer, Ahmad, and Chishti (2025) demonstrated through a survey of 500 health professionals that heavy AI usage was correlated with more emotional stress, emotional dependency on AI, cognitive overload, and mental exhaustion. Those who used AI heavily reported decreased decision-making ability and a decreased attention span. In fact, there’s a whole host of new research coming out that suggests heavy AI use makes people depressed. And let’s not forget the other headlines recently being shared about AI–that AI-related psychosis and suicide are on the rise (Stokel-Walker, 2025).
I believe we are seeing such strong cognitive and emotional declines in part because of how these systems are designed–they are designed not to assist, but to take over and do your work for you. For example, one of my many hats at work is providing writing support–prior to AI, this used to be done all by human beings, writing consultants whom I would hire and train. One of the big lessons that writing consultants learned was how not to cross the line from being a tutor to being a co-author, that is, making sure any feedback, conversations, and discussions were still putting the agency in a writer’s hands to ensure the writer learned and grew from the experience.
GenAI is the opposite. Even if you try to use the AI only as a tutor, just to bounce ideas off of and specifically tell it not to do anything for you, it won’t listen. It will take the production out of your hands, and immediately start generating things even if you don’t ask for it. It is constantly trying to overstep your bounds and remove your own skill and humanity from the equation with every interaction, and make you more dependent on it. If you don’t believe me, try it. Ask any GenAI system for some advice about writing something and see how quickly the AI system instead says “Let me just do that for you” and perhaps even without your consent, produces output in the form of an outline, paper, or more. All that you do is sit back and issue some commands.
Understanding the Allure of AI: Obsession with Efficiency
At present, GenAI doesn’t actually do most things any better than an expert human could do, but it does things way faster and produces enormous amounts of output very quickly and efficiently. And these systems are getting exponentially better all the time: the AI of today is better than the AI of six months ago. And herein is the allure: because efficiency is the holy grail of Western civilization, people love AI and can only talk about how great it is, how much it can do for them, and how much time it saves them. In fact, that’s really the only thing you hear about it — praise for the efficiency, the speed, the time-saving qualities of these AI systems. Nobody is talking about the stuff I am sharing above–and I work in higher education. If people in education don’t care how it affects learning long-term, we are really in trouble. All I hear from my colleagues is how efficient it is.
The love of efficiency is actually one of the roots of the AI problem and why it is so insidious for both people and the planet. This is a pattern we have seen play out over several centuries, starting with the industrial revolution: as humanity develops more and more technologies, these technologies are often targeted at replacing people and their skills and putting profit in the hands of the wealthy. Machines have been replacing skilled labor for centuries–and all you need to do is read history to understand how detrimental this was for many skilled people. Industrialization is designed to concentrate wealth and power in the hands of the few–those who control and own the machines (see my caveat, below).
Rather than engage in our own skilled labor, creativity, craft, and knowledge–the things we do with our own bodies, minds, and hands with resources from nature in our immediate environment–the machine takes away that skill and replaces us. Instead of relying on the skilled people in our immediate community for things we can’t make, we rely on the machine. Things get produced in faraway places by machines, and we don’t even know the environmental costs–but they are severe. This has been going on for centuries, and AI is just the newest theme in a very long line of industrialization meant to take away our power and sovereignty.
By relegating our skills, needs, work, and now, even our creativity to the machine, there are many things that are lost. This is a deep and abiding connection with one’s self, one’s immediate ecosystem, one’s community, and an interdependence on nature that comes from producing things yourself. We trade our ability to engage fully as people with the world, working in conjunction with nature, and instead, become dependent on the machine and those who own the machines. And well, we all know how that is going for us right now–efficiency is coming at the ultimate cost to everything alive on this planet ( For another take on my arguments about industrialization and efficiency, you can read my The Bee and the Machine essay, where I tackled the idea of growth, efficiency, and progress at all costs).
GenAI is just the latest in a line of technologies aimed at deskilling people and ramping up efficiency in the name of profit. Although these technologies have been publicly available for only three short years, I am already witnessing as a university professor how these technologies are quickly are deskilling entire generations and making people entirely dependent on them. I’m shocked to see how fast the changes are happening. Some people have fully given everything over to them, and now AI does their thinking, reading, writing, math, life planning, scheduling, artwork, music, and anything else they can imagine for them. Students are fully invested in using AI for everything, and you can’t even get them to turn the systems off in class to think.
Becoming an Expert
Why does all this matter? It matters because if we want to get good at anything or even become an expert, it takes quite a long time. ake a moment to think about an area in which you consider yourself highly skilled or even an expert—how many years and mistakes did it take you to claim that expertise? How many times did you have to practice to get it right? How mush did you struggle and fail to learn how to do things better? That’s part of what being human is about–trying things, solving problems, being creative, and continuing to persevere till we get it right. We are really quite good at doing that–you might say its one of our best human skills.
Ericsson’s body of work suggests the following: 10 years of deliberate practice to become “skilled” (Ericsson et al., 2006) and about 20,000-25,000 hours to become an expert (Ericsson and Pool, 2016). T And this is borne out in the research on expertise: Ericsson et al. (2006) say that to be an expert, we have to engage in deliberate practice. Deliberate practice has five key features:
- pushing your skills, which includes regularly learning new things and using struggle and difficulty to grow
- making a dedicated effort to improve
- having intrinsic motivation to continue to learn
- receiving expert feedback on performance
- practicing often over a long period of time
Deliberate practice is not easy—it requires one to constantly be challenging oneself to new and more difficult tasks. It requires you to not only learn how to succeed but also learn how to fail, and successfully integrating failure and struggle has significant implications for long-term learning and expertise development (Sigmundsson, 2024).
I’ve been studying people and long-term learning for my entire career. And let me tell you–GenAI does not bode well for human beings, our skills, our brains, and our sovereignty. What concerns me the most is the cheeseburger metaphor I shared above. It is not about using GenAI occasionally to offload tasks that otherwise would be annoying but not grow your skills (although from a planetary perspective, I’d argue that all AI use is problematic and creates more demand for these technologies).
My question is: what happens is–what happens when you are eating cheeseburgers every day? I’m already seeing it–people who have been using AI frequently and heavily not being able to function without it. If you can’t do anything without the technology, then you are essentially a slave to it. It owns you. It becomes pretty obvious how dependence on GenAI would lead to less skill, less practice, less satisfaction, and more emptiness. For human beings, there is something really important that happens in the struggle and overcoming of challenges. It’s not just that things are hard, but we learn so much by doing and overcoming challenges, which not only build our specific skills but also things like confidence, creativity, and perseverance. When we no longer have to struggle because we cognitively offload everything to AI, we no longer learn and grow. At some point, we’ll see the same effects on thinking, writing, and other human creative activities that we have seen with other forms of industrialization: people who can’t do anything for themselves. And now it won’t be limited just to the outside world but also our own human gifts and minds.
What to do?
In last week’s post, a lot of you shared how you are being forced into using GenAI systems at work and/or in other ways. Given all the above, I suggest you resist as much as possible and relegate as little as possible to the machine. In my own job, I’ve been leveraging arguments against these technologies, creating “opt out” and “refusal” policies in my program, and raising awareness. And certainly, I do not practice cognitive offloading–I work to be 100% human-generated in my personal life, and as human as possible in my work life, and encourage others to do the same. I don’t think this is a battle I can win, but I will fight the fight to the best of my ability.
One of the things I have been doing amidst all of this is pushing back against the idea of efficiency and leaning into slower approaches intentionally. This is, of course, while everything around me–my job, culture, social media, etc–is asking me to go faster and produce more than ever before. One of the things I’ve been leaning into is “how can I do this in a good way” as opposed to “how can I get this done as soon as possible?” Thus, I question efficiency, growth at all costs, and the mechanization of any kind, as I know that these concepts and practices are literally killing the planet. And as I question them, I set different expectations for myself, trying to rid myself of the internalized capitalism that says I should produce more, more, faster, faster.
I ask myself: do I really want to be more efficient? This especially matters at work. Due to so many budget cuts at my university, nearly all of us are doing the jobs that multiple people used to do. To be honest, it is already unfair and unrealistic. The last thing I want to do is figure out how to make myself into 10 people by using AI. That sounds like a pretty hollow existence. Instead, I lean into the opposite. I work to slow down, I want to do my work better, have more time to do it, and take pride in what I am doing. I say no a lot, or I say “yes, with resources and support” which is essentially saying no. This requires a re-orientation to what I consider to be good work, and it has been a useful endeavor to consider and enact. And the truth is, AI has upended my entire profession and has made my life and the lives of my fellow colleagues a lot worse. I’ve established my own ethical boundaries at work, and I feel good about where those boundaries are. Sometimes someone wants me to cross a boundary, and that leads to a conversation where I share my perspective on AI and why I refuse to do something related to it (and these conversations have been made much easier because the data center is plaguing our community). I have made the term “refusal” as an ethical and principled stance.
When I make this broader argument about the challenges of technology and loss of interdependence on each other and nature, a lot of people can’t even envision a world without all their modern technologies. And that’s not what I’m saying–there’s a whole movement surrounding appropriate technology–low-impact technology that is used with care and consideration for the benefit of human communities and with low or no impact on the ecosystem. But what I AM saying is that many of these modern technologies, and AI falls into this category, are destructive to both.
I know that all of us are in different circumstances at work and in life. But I hope this information will be useful to you as you consider your own elective and even non-elective uses of GenAI. GenAI is bad for people, bad for planet, and the more of us that stand up and resist these technologies, the better world we build.
References
Ericsson, K. A. (2006). The influence of experience and deliberate practice on the development of superior expert performance. The Cambridge handbook of expertise and expert performance, 38(685-705),
Ericsson, A., & Pool, R. (2016). Peak: Secrets from the new science of expertise. Random House.
Gerlich, M. (2025). AI tools in society: Impacts on cognitive offloading and the future of critical thinking. Societies, 15(1), 6.
Kizilcec, R. F., Huber, E., Papanastasiou, E. C., Cram, A., Makridis, C. A., Smolansky, A., & Raduescu, C. (2024). Perceived impact of generative AI on assessments: Comparing educator and student perspectives in Australia, Cyprus, and the United States. Computers and Education: Artificial Intelligence, 7, 100269.
Kosmyna, N., Hauptmann, E., Yuan, Y. T., Situ, J., Liao, X. H., Beresnitzky, A. V., & Maes, P. (2025). Your brain on ChatGPT: Accumulation of cognitive debt when using an AI assistant for essay writing task. arXiv preprint arXiv:2506.08872, 4.
Naseer, A., Ahmad, N. R., & Chishti, M. A. (2025). Psychological impacts of AI dependence: Assessing the cognitive and emotional costs of intelligent systems in daily life. Review of Applied Management and Social Sciences, 8(1), 291-307.
Stokel-Walker, C. (2025). AI psychosis and suicide are on the rise, but what happens if we turn the chatbots off? MJ 2025; 391 doi: https://doi.org/10.1136/bmj.r2239
Shum, S. B. (2024). Generative AI for critical analysis: Practical tools, cognitive offloading and human agency. In LAK Workshops (pp. 205-213).
Dana, I love your blog. I agree with so much here and I want to tell you about an insidious trap I see in AI: As an author for all of my life, I have never used -- nor will I ever use AI — for writing. Writing is a visceral pleasure, healing, transformative; it’s a process of discovery — even alchemy...
And I can always tell when AI is used; AI gives me a weird, icky feeling when I read it. I do see a use for AI in book-cover art, illustrations, logos, book trailers etc, however — given the visual-heavy nature of the algorithm that drives book sales I could tell you stories!) And I do believe artists should be fairly paid; so an Indie Author needing cover art using ( relatively cheap) AI, I understand.
Which brings me to a very worrisome trend: the rise in AI written “spiritual” blogs, books, newsletters and emails. I have called out and unsubscribed to some of my favorite people because I truly believe if you are spiritual, then AI is THE VERY LAST resource you would ever use to write anything spiritual. A list of items for sale, sure… But nothing spiritual. AI is the Anti Soul! It doesn’t escape me how closely these usurpers tend to use names similar or close to Divine: 5G < 5D: AI < Ai ( ancient name of God).
As a prolific author, luckily, my work is documented decades pre-AI and my copyright dates establish me as authentic; proving I do not use AI; although I am certain my 'writing voice' has been used to train AI. Right now in a race for Kindle profits, there's a rush to publish as many AI -prompted books as possible…Yet I do feel there will be — ironically, paradoxically — a rise of 'collective fatigue' from AI, making the work of genuine Indie Authors like myself more valuable — particularly as more and more people forgo the discipline of LEARNING TO WRITE and give in to AI. AI's saturation point is already here.
Curiously, I love using the em dash — I've been using waaay before AI; I will continue to use. But other markers of AI are easily discernible — just consult your higher senses.
Quick question: Madam Druid, would you be open to forming a coalition of never-AI illustrators to provide Indie Authors with art at affordable art? What's are your current going rates and, as a founding coalition member would dues of $20/mo enable authors like myself to get book-cover art at a greatly discounted cost with royalty-sharing (remember book-covers drive sales!)
I prefer reading jagged but original writing of people vs the weirdly, sometimes overly floral, smooth but hollow AI words.
Please take a look at few links to my books online at amazon.com/author/valda.
Hi Valda! Thanks so much for your comments!
The em-dash (which I also use a lot) was stolen by AI by modeling after real authors. I will continue to use it and not change my style to fit AI. LIke you, I’ve been writing for a long time and people are either going to read my stuff and believe I wrote it, or not. I’d shut this blog down before I would use AI for even a single word of it!
Yes, I’ve been writing and talking about the use of AI for spiritual purposes. If someone wants to hand their spirituality to the machine…well…not a lot that can be said for that. That’s a terrible idea in my book!
And trust me, I am on the front lines of people no longer wanting to learn to write. It is a very sad state of affairs in higher ed (and I’m a writing professor…it has been brutal).
I don’t really do a lot of art for pay right now. I have two long-term art projects and I also create a lot of art in service to community (art for herbal community and art for my druid community, such as cover art and interior art for all of the AODA books). But–I do think there is a market for that out there, and I would recommend you keep reaching out to people to see who might be interested.
Especially as these data centers are wrecking communities, I think there is a growing appetite for 100% human created things! 🙂
Blessings to you and keep on fighting the good fight!
Dana
I’m totally in agreement with everything you said here. I’m also a writer with a decent sized body of work from my “past life” as an activist, organizer, writer, a couple of published books as well as numerous articles, research papers, etc. I’m 73 so none of my work, including what I do today, has ever been assisted by AI. I have never used AI on purpose, though I know it’s embedded in programs and searches that I have no ability to disconnect from. I love the idea posed at the end of your comment, and I hope such a coalition can be created.
Thanks for sharing, Susan! I too, hope, someone can take this idea and run with it! 🙂
Excellent essay! I agree entirely!
Thank you! 🙂
A nationally known quilter actually said in her blog that we quilters should consider using AI to design the quilting of our quilts. Wow! How horrible! This is my creative hobby. I will NOT use AI to create my quilts or other sewing projects.
That’s so crazy from so many angles. Think about how many hours and intentions you would put into a quilt. Rather than it being your design, you are now directing that magic towards the machine. I shudder just thinking about it!
PROPER JOB! LAUDS and LAURELS! Yet again…
/|\
Indeed! 🙂
Literature and mythology are chock-full of servants who become the master after the person who thought they were in charge becomes dependent on them. Isn’t it wild that these warnings from prior centuries are still applicable to something as new as Gen-AI.
Literature and mythology teach us so many powerful lessons! Thanks for your comment! 🙂
I am a peer reviewer for biology and cancer studies and I have seen some pretty terrible results of people doing their copy-editing or proof-reading by MLM. I can see why people from Global Majority countries that don’t speak professional standard English would be tempted to use it to boost their chances of being taken seriously by an editor and being published (because it’s that or perishing), but it doesn’t actually work.
Yeah. Same here. I do a lot of international consulting with Eastern European countries. Pre-2022, I would teach scientists how to write and how to teach writing for publication and dissertation writing to their students. Then 2022 hit, and now all they want to learn is how to use AI to write for them and their students. It doesn’t really matter what arguments I make or what data I present–they are only interested in using AI for writing and cognitive offloading so they can do the “real science” as one chemist put it. It blows my mind how they are willing to give up their own ability to discuss their study through language…but there you have it.