Why I Let AI Help Me Think — But Never Think for Me

Every technological leap generates its share of moral panic, and AI was never going to be an exception. An article in The Washington Post a few days ago, “Meet the people who dare to say no to artificial intelligence”, is about the growing number of students, artists, academics and freelancers who are rejecting the use of AI in the name of creative purity, fear of bias or simply in defense of human work. All very noble, but surely this is more a form of denial than resistance.

Refusing to use a technology that is already part of education, communication and work is, in the end, simply an attempt to isolate oneself from the future. It is not a matter of blindly believing in AI or embracing it wholesale, but rather one of understanding and using it judiciously. Because those who say “no” outright today will end up less prepared to live in an environment where it will be a part of everyday life.

My position on AI is neither one of naïve enthusiasm or rejection: it is one of selective and strategic use. I use it when it improves my efficiency, but I keep it well away from what I consider essential to my identity. I’m not an illustrator, so therefore I’m not worried about using AI to generate images to accompany my articles. I used to depend on image banks, and from time to time an artist would appear claiming rights or asking me for proof of the provenance of the illustrations I had used years ago. Now, with a generative model, the process is fluid, immediate, and frictionless: no one claims anything, and my texts are better illustrated (although not all my readers approve).

I also have no problem asking a language model to check my typos, summarize long documents for me or suggest counterpoints to an article I’m writing. I do it often. Thanks to that, I make fewer mistakes, save time and I can enrich my arguments. I don’t always, but I do it when I see that it makes sense. And it does not impoverish my writing, quite the opposite.

Could I go further and ask an AI model to write my articles? Yes of course. It would be enough to train one with my texts, based on more than twenty-two years of publications at least daily, and it would obtain a perfect copy of my style, my structures and my turns of language. It would be easy, cheap and, probably, the result would be indistinguishable from the original, or who knows, even better. But I have no intention of doing so.

Why? It’s not about ethics, but because writing is what defines me. Writing is my way of learning, understanding and remembering. If I delegate that process to a machine, I am sabotaging the mechanism that sustains my own knowledge. My articles are more than opinion pieces: they are my method of thinking. In fact, my opinion is what matters least. I’ve said it thousands of times: pay less attention to my opinion, and more to the links. I write to order ideas, to fix them in my memory, to be able to explain them and discuss them later in class or in a conference. If I outsourced my writing, I wouldn’t be able to defend my arguments, to answer a question in a debate or remember why I thought what I wrote. AI might do better in form, but never in purpose.

That relationship with technology is what many seem not to understand. The problem is not AI: the problem is not knowing what we are using it for. The Brennan Center for Justice recently warned that AI opt-out policies create more inequality because those who refuse to learn how it works tend to be marginalized professionally. Similarly, the MIT Technology Review says avoiding AI doesn’t preserve critical thinking; instead it reduces cognitive capacity, in contrast to those who learn to use it judiciously.

Empirical research confirms this. According to a June 2024 study, over-reliance on conversational systems can erode cognitive abilities in students, but the strategic use of those tools, as a support rather than a substitute, strengthens understanding and creativity. In the future, all hard skills learning will be taught by agent systems and generative models, which will explain them to you as many times as necessary, 24/7, and with the format that best suits your way of learning. So the problem, as always, is not the tool, but how it is used.

AI doesn’t replace my thinking, it amplifies it. It helps me illustrate, proofread, research or prepare material, but not write. Because writing is what keeps me alive intellectually. The key is understanding which tasks to delegate and which to preserve. What some researchers call cognitive offloading, ceding part of the mental effort to automatic systems, can be useful as long as it does not imply giving up the competencies that define us. Refusing to use AI to write does not make me an AI denier. I use it every day, but with clear limits. I do not use it to replace myself in what I consider to be an essential part of my intellectual work. I use AI; I don’t let it use me.

In the final analysis, rejecting AI is no different to rejecting books. The tools may change over the years, but the criteria for using them are still human. As Jan-Christophe Heilinger warns in his essay on the ethics of AI, the discussion should not focus on whether or not to use it, but on how to retain human agency and responsibility within an automated ecosystem.

And yes, there are those who come to my page and, without knowing me at all, say that my texts “seem to be written by AI”. It’s funny: I’ve been posting daily for more than twenty-two years, even before ChatGPT existed, and I still write word for word. Even so, there is always some idiot willing to doubt one’s integrity. I suppose it is part of the times we live in: a world in which many prefer to suspect rather than read, and deny rather than understand. But what are we going to do, there have been imbeciles and rude people all our lives and, unfortunately, there will continue to be.

The key is not to say “yes” or “no” to technology, but deciding what you use it for. I’m not going to use AI to replace my thinking, instead, it augments it. It helps me illustrate, proofread, research, or prepare material, but not write. Because writing is what keeps me intellectually alive. I suppose that others more intelligent than me get by just by reading. But I need to write down what I learn.

In conclusion, literacy in the 21st century is not about avoiding technology, but about learning to use it in a way that protects what makes you unique. aying “no” on principle is the easy way to avoid thinking. It’s much harder to build an intelligent relationship with technology. Anyone who doesn’t understand that will end up discovering, too late, that the world kept moving forward while they looked the other way. And then, the problem will no longer be AI, but the decision not to have learned how to live with it.

—

This post was previously published on Enrique Dans’ blog.

***