I have a small confession. This text? It is not the first draft. Not the second either. At some point, I lost count, but I think the current version is somewhere around the sixth or seventh.
Now, I could tell you that every word here was carefully weighed, that every sentence went through a painful process of deletion and rewriting. And that would be true. But it would also be only half the story.
The other half is that alongside this document, I have another window open. A window with a digital creature that does not get tired, does not complain and does not ask, “Why do you need another version? The third one was great.” It is just there, waiting for instructions.
Suddenly ,we have a diligent, brilliant personal assistant who does not sleep at night. We type a few words, sometimes not even a full sentence, just a vague idea, and boom, we become the best version of ourselves. Faster, more efficient, more creative.
But the real question, the one almost no one asks, is this: could tools that generate words and phrasing for us actually make us more mediocre?
Not dumber, God forbid. Mediocre. People who stop straining to find the precise word, because someone else will find it for us. A tool that can write, organize, throw out ideas and free up our time. But in practice, a different pattern begins to form: we quietly slide into comfortable mediocrity. Instead of leveling up, we are steadily pulled toward the middle.
For years, we heard the phrase “think outside the box” in the context of creativity and innovation. And now we have reached the moment when humanity has invented the most sophisticated box in history: artificial intelligence.
The problem is that speed blinds us. The model spits out an answer in seconds, and we, creatures who interpret effort as value, automatically assume something creative has happened. But the truth is less dazzling. AI is still a kind of massive archive containing almost everything humans have ever written and created.
So what happens to the brain when we let AI write for us?
Researchers at MIT wanted to test exactly that. They took more than 50 students, hooked electrodes to their heads and gave them a simple task: write an essay on a question like “What is happiness?”
The students were divided into three groups. The first wrote on their own, from their heads. The second was allowed to search on Google. The third was given access to ChatGPT.
While the researchers themselves cautioned that the findings are preliminary and far from definitive, the results painted a surprising and clear picture.
The group that worked with ChatGPT showed significantly lower brain activity than the others. Less effort, less thinking. When participants were asked to quote a sentence from their own essay, 80% of the AI users could not remember anything. They wrote it, but it did not stay with them. While the first two groups showed a range of opinions and ideas, the essays produced by the AI group were nearly identical twins. The same phrasing, the same polished mediocrity.
How mediocrity is created
Put simply, the technology smooths out the sharp edges, the unusual phrasing, the things that may have been a bit clumsy but were yours. To understand why this happens, we need to zoom in. These models were trained on enormous amounts of text: books, articles, Instagram posts, recipes, almost everything. And when they respond, they essentially produce a weighted average of what they have seen, choosing the safest and most predictable path.
You can argue that writers have always relied on aids, and that is true. But in the past, those aids were only a small part of the process. Today, a language model like ChatGPT can do almost everything, from writing emails to drafting legal documents.
And that may be the heart of the issue. The wild capabilities of AI are no less important than our human ability to activate and direct it. The tool can produce endless variations, but it is not the one that decides to break routine or step outside the box. We decide.
How not to become average
It is possible to significantly improve the quality of the results, to rise above mediocrity and flat answers. From experience, these models can actually strengthen original thinking if they are not used merely as a convenient solution.
One trick that works well for me is to deliberately introduce friction. The natural tendency is to tell the model, “Improve this” or “Write an opening.” Easy and fast. But then our brain goes to sleep. Instead, try throwing it strange demands. Ask it to explain a complex concept without using the word “it.” The odd constraint forces the model off its usual tracks and into less predictable territory.
Another approach I like is to ask the model to map the danger zone in advance. Before I start writing, I ask: “What are the five most predictable things people say about this topic?” What comes back is essentially a list of clichés to avoid. From that point on, I have a blacklist.
Try this too: mix domains. The model likes to stay within the genre you give it. If it is business, it stays business. If it is technical, it stays technical. But if you ask it to explain a technical process as if it were a soup recipe, suddenly it has to work harder. The strange combination pushes it out of its comfort zone. Try it. I promise you will discover new worlds there.
Keren ShaharYou can even reverse the roles. Instead of asking, “Write this for me,” try, “Find all the holes in what I wrote, and do not be gentle.” Suddenly, you have someone who is not trying to please you, but to annoy you. And that is exactly what forces you to sharpen what you wanted to say.
The model is a tool, humans create meaning
Yes, a person equipped with a smart tool can truly go further than a person without one. That is a fact. But the tool itself, no matter how sophisticated, is completely passive. Artificial intelligence is a powerful tool for generating variations on what already exists. It can replicate and optimize “inside the box” perfectly.
But if you are looking for the spark, the deviation from routine, do not expect the tool to do it for you. The box will remain a box. The responsibility to step outside it and create something genuinely new, as always, remains with us.