The Real Cost of AI Isn't Copyright

It Could Be the Decline of Thinking

Anthropic found itself at the center of another debate about artificial intelligence and intellectual property two weeks ago. The company will have to pay $1.5 billion for using copyrighted books to train its AI models. This is a reminder that these systems were built on the work and intellectual property of writers, journalists, researchers, and publishers who spent years creating their original ideas.

This copyright issue is important if you are a creator. If AI companies profit from other people’s work, then the original authors and artists deserve compensation.

But I think there’s a more profound issue hiding underneath the headlines and legal wrangling.

If AI can summarize every book, rewrite every article, answer every question, and even generate new prose in seconds, will people still choose to read and think carefully and deeply?

And if they don’t, what happens to our ability to think?

Rose Horowitch’s recent Atlantic essay, The End of Reading Is Here, makes a strong argument that we are entering what she calls a “postliterate” era. This era is not one where people can’t read deep and intricate content, but one where they choose not to do so. Humans may be consuming more words than ever through texts, emails, social posts, and captions, yet they’re spending less time with complex writing, long form books, and articles. The result isn’t illiteracy or short attention span; it’s the gradual erosion of thought, comprehension, synthesis, and most importantly, deep reasoning.

Horowitch writes,

“[R]eading is not simply a way of acquiring information; it is a way of developing the capacity for sustained attention, inference, and judgment.”

Those are muscles that weaken when they aren’t exercised.

As she points out, people are probably reading more words than ever before, but what has changed is what they read, where they read, the format of the prose, and how much they absorb. We are bombarded with emails, text messages, posts, Reddit threads, and Instagram captions. That explosion of textual fragments has come at the expense of sustained attention which is required for longer works that convey rich, more complicated, and nuanced ideas.

This post also appeared in The Systems Leader Newsletter.

Reading a thousand tweets is not the same as reading one thoughtful book, in the same way watching hundreds of thirty-second videos isn’t equivalent to wrestling with a difficult argument over several chapters. Our brains aren’t simply consuming information. Rather, they’re being trained in the format which that information is delivered.

What is ironic is that AI is emerging just as our attention spans are already weakened by smartphones, short form content, and social media. Horowitch states that generative AI may represent the first technology that doesn’t merely distract us from writing, it threatens the very practice of writing itself.

She points to research suggesting that when people rely on AI to compose their work,

“[T]hey may produce acceptable writing while retaining less of the knowledge and doing less of the thinking that writing itself requires.”

Writing has never been just a way to record thoughts. Writing is how we discover ideas, dissect them, work through challenging issues, and deliver quality work.

Anyone who has written a book, or even a long-form Substack or blog, knows that you don’t start with a perfectly organized argument. You begin with fragments of ideas and a tentative notion of where you are headed. Writing forces you to clarify your thinking, identify contradictions, eliminate weak logic, and connect thoughts that weren’t previously aligned.

The process changes the author as much as it produces the final product.

That’s because writing well forces critical thinking. Horowitch notes that many writers discover what they have to say only through the act of wrestling with language on the page or screen and struggling with their ideas. AI can generate polished sentences but it can’t replicate the sometimes painful intellectual journey that produced them.

If AI does the work for us, we may save time but we risk losing one of the primary mechanisms through which humans develop original thought. As Horowitch notes, early studies already suggest that while AI makes writing easier, frequent reliance on it may come at the expense of long-term learning.

One researcher Horowitch cites worries that convenience may come with an invisible tradeoff: if machines perform the heavy lifting our brains should be doing, our capacity for analysis and memory may gradually atrophy.

I’ve seen first-hand what AI does to our students. The LLMs are getting better, the students are improving at using these tools, but do our young men and women deeply understand what they are handing in?

AI gives us intelligence, but it doesn’t give us wisdom.

Is Writing Still Worth It?

That raises an uncomfortable question:

Is writing books or any long-form content still worth it?

This hits close to home as I begin the task of writing my third book.

Commercially, publishing has never been easy or always profitable, and AI is already flooding the market with competent, inexpensive content. Horowitch notes that AI has increased the number of books being published while simultaneously making polished writing easier to produce.

But I think that’s the wrong way to frame the question. The value of writing a book or essay has never been limited to selling copies or making money off of the process.

Books and long-form writing establish deeper understanding of the human condition. They organize our experiences as a writer and reader. They preserve ideas that can influence people for years after they’re written. They force authors to think more clearly than they otherwise would. And they force readers to contemplate that which they have not previously considered.

Already short-form content is dominating awareness.

Long-form content will shape what exactly? Deeper thinking?

Maybe?

The challenge for leaders is understanding that these formats serve different purposes. A short video may capture attention, but a thoughtful article builds clarity.

A well-written book can build a foundation.

The companies building AI must compensate the people whose work made these systems possible. Copyright law matters. That’s an important legal and ethical conversation.

But an even more important one is this:

If we have entered the postliterate world, how will humans develop and wrestle with overtly complex issues? And how will timeless learnings survive over time?

And if they don’t, does it even matter?