AI can generate almost anything. But somewhere along the way, we started confusing the ability to produce with the ability to create.
There is something almost absurd about the age we are living in
.
You type a few words. Press Enter. And suddenly you have an article, a painting, a business plan, maybe even an application.
A few seconds later, you are a writer. A designer. A developer. At least according to the enthusiasm surrounding generative AI.
Really?
I use AI. I experiment with it. I am fascinated by what it can do, and I have no interest in defending a world in which creativity belongs exclusively to people with the right qualifications.
But I think we are making a fundamental mistake.
We have confused access with ability.
And the more impressive AI becomes, the more dangerous that confusion gets.
The Biggest Lie of the AI Era
I keep coming back to one sentence:
The biggest lie of the AI era? Believing that anyone who knows how to hit ‘enter’ will become an author, a designer, or a thinker.
— BJ Beyond
It is not an attack on technology.
It is an attack on an illusion.
Owning a camera doesn’t make you a photographer. Having a piano in your living room doesn’t make you a musician. And opening ChatGPT doesn’t automatically make you a thinker.
Of course, you can learn.
You can become very good.
AI might even help you get there faster.
But there is a difference between becoming capable of doing something and receiving a finished result from a machine.
That difference matters.
The Problem Isn’t That AI Creates. It’s That We Stop Thinking.
I don’t believe the real danger is that machines will become creative.
Perhaps they already are, depending on how we define creativity.
The more immediate danger is that we will become less demanding of ourselves.
A beautifully written paragraph appears on the screen. We assume it makes sense.
An image looks extraordinary. We assume it expresses something meaningful.
An application works. We assume its underlying code is reliable.
Why question something that looks so convincing?
Because appearance is not understanding.
A machine can produce an argument without the person requesting it having examined its foundations.
It can generate code without the person publishing it understanding its vulnerabilities.
It can produce a visual composition without anyone deciding what that composition is supposed to communicate.
And yet the result can look professional.
That is what makes this moment different.
The distance between competence and the appearance of competence has become remarkably small.
Science Is Beginning to Show the Paradox
This is not just a philosophical argument.
In 2024, Anil Doshi and Oliver Hauser published a study in Science Advances examining how generative AI affects creative writing.
Their findings were fascinating.
People who received AI-generated story ideas produced stories rated more favorably on creativity and other measures, particularly when their initial creative ability was lower.
That’s good news.
But there was another finding: AI-assisted stories were more similar to one another.
Individual performance improved. Collective novelty declined.
The original research is available here: Generative AI enhances individual creativity but reduces the collective diversity of novel content.
Think about what that means.
We may be helping individuals produce better work while simultaneously making the overall landscape of ideas less diverse.
Better results. Fewer differences.
That’s a paradox worth discussing.
And the evidence did not end there.
In May 2025, Lennart Meincke, Gideon Nave and Christian Terwiesch published ChatGPT decreases idea diversity in brainstorming in Nature Human Behaviour.
Their contribution reexamined evidence from earlier brainstorming experiments and highlighted how AI assistance can narrow the variety of ideas generated by groups.
The lesson is not that AI always makes people less creative.
It is that improving individual ideas and increasing the diversity of ideas are not necessarily the same thing.
Then came another interesting contribution.
In a 2025 paper published in Computers in Human Behavior: Artificial Humans, Kibum Moon, Adam E. Green and Kostadin Kushlev examined 2,200 college admissions essays across three preregistered studies.
Their conclusion deserves attention.
Human-written essays introduced new ideas at a higher rate than essays produced by GPT-4 in the conditions they examined. That difference persisted even after attempts to improve the diversity of the AI-generated material.
You can read the study here: Homogenizing effect of large language models on creative diversity.
None of these studies proves that AI is incapable of originality. Nor do they tell us what future models will achieve.
But together, they raise an uncomfortable question.
What happens when millions of people begin outsourcing not only the execution of their ideas, but the ideas themselves?
The World Doesn’t Need Another Million Identical Voices
Look around.
We are producing more content than ever.
More articles. More images. More videos. More opinions.
Everyone has something to publish.
But how much of it actually says something?
I am not interested in the romantic idea that everything made by human hands is automatically better.
Humans produce mediocre work every day.
And AI can produce remarkable work.
The distinction I care about is somewhere else.
Why did you create it?
What did you want to express?
What did you reject before arriving at the final version?
Do you understand what you are publishing?
Can you defend it?
Those questions matter more to me than whether someone used a pencil, Photoshop, or a language model.
The tool is not the enemy.
Intellectual passivity is.
Pressing Enter Is the Beginning. Not the Achievement.
Imagine two people generating exactly the same image.
The first accepts the initial result because it looks impressive.
The second questions it.
Why this composition? Why these symbols? What is missing? Does it express the intended idea or merely resemble something already familiar?
The second person changes direction, rejects alternatives, makes choices.
They may use AI throughout the process.
But they are not simply receiving an output.
They are exercising judgment.
And judgment is not a minor detail.
It is the difference between letting a system decide what is acceptable and deciding for yourself what is worth creating.
Of course, this distinction is not absolute. A brilliant idea can emerge immediately, while hours of revisions can produce something ordinary.
But the ability to evaluate a result remains essential.
That’s where I believe the Human Edge begins.
Not in the refusal to use technology.
Not in nostalgia.
And certainly not in pretending that humans are always superior.
It begins in the willingness to remain intellectually responsible for what we produce.
The Human Edge
I have spent considerable time exploring the relationship between human identity and artificial intelligence.
What interests me is not simply how much AI can do.
It is what happens to us when it can do almost everything we ask.
Do we become more curious?
More capable?
More independent?
Or do we gradually accept whatever arrives on the screen because questioning it takes longer?
These are different futures.
And I don’t think technology alone will decide which one we choose.
There is a great deal of talk about prompts, automation and productivity.
All legitimate subjects.
But I would like to see more discussion about judgment.
About knowing when an answer is wrong.
About recognizing when an image is empty despite being beautiful.
About refusing to publish a convincing argument that we have not bothered to understand.
AI may become extraordinarily good at producing answers.
That makes our ability to question them more valuable, not less.
The Real Opportunity
I don’t want fewer people creating.
I want more.
I want someone who has never written a line of code to discover that they can build something.
I want a person who struggles to express an idea to find the words.
I want artists to experiment with techniques that previously required resources they could never afford.
That is the promise of AI.
And it is a powerful one.
But I don’t want us to confuse the possibility of becoming something with the illusion that we have already become it.
A beginner using AI to learn programming is not pretending.
A writer using AI to challenge their assumptions is not surrendering authorship.
A designer using AI to explore possibilities is not abandoning creativity.
The problem begins when the work stops at the first convincing result, and we declare the process complete.
Because technology can remove barriers.
It cannot guarantee understanding.
One Last Question
Perhaps one day machines will write better novels than humans, produce more compelling art, and develop ideas we would never have imagined.
I don’t consider that impossible.
But even in that world, one question will remain.
What do we choose to do with those capabilities?
We can use AI to expand human thought.
Or we can use it as a substitute for thought.
The difference may not always be visible in the finished product.
It will be visible in the decisions we remain capable of making.
So yes, press Enter.
Create. Experiment. Build things that were impossible yesterday.
But don’t confuse the button with the mind behind it.
The future doesn’t need more people who know how to generate an answer.
It needs people who still know how to question one.
Sources and Further Reading
- Doshi, A. R., & Hauser, O. P. (2024). Generative AI enhances individual creativity but reduces the collective diversity of novel content. Science Advances, 10(28).
- Meincke, L., Nave, G., & Terwiesch, C. (2025). ChatGPT decreases idea diversity in brainstorming. Nature Human Behaviour, 9, 1107–1109.
- Moon, K., Green, A. E., & Kushlev, K. (2025). Homogenizing effect of large language models (LLMs) on creative diversity: An empirical comparison of human and ChatGPT writing. Computers in Human Behavior: Artificial Humans, 6, 100207.
BJ Beyond explores the Human Edge: originality, identity, critical thinking and the relationship between humanity and artificial intelligence.
Website: https://bjbeyond.it/
Editorial note: This article was developed from BJ Beyond’s original thesis with AI-assisted drafting and editorial support. Final editorial responsibility rests with the author.
