What AI Is Taking—or What We’re Giving Away

We are reading less. Our attention is divided—subdivided—more than a coastal farm in Ireland. So much so that we no longer see the fields, only the walls. This fragmentation predates AI, but AI isn’t a remedy unless we shape it into one.

Take the choice between spending 20 hours with Ada Palmer’s Inventing the Renaissance or the same amount of time querying Perplexity AI on the topic. AI can match Jeopardy performance, fulfill curiosity at scale, and even generate visuals on demand. It often makes me feel smarter. But it can’t give me Palmer’s voice. Her work doesn’t just inform. It provokes, reframes, and lingers. It made me want to revisit Florence. And when Sam Altman likened AI to the Renaissance rather than the Industrial Revolution, I didn’t have to guess what Palmer might say. (Spoiler: It wouldn’t go well for Mr. Altman.)

We Imagine Less

We know watching a movie isn’t the same as reading a book. But what, exactly, do we lose?

When my 11-year-old daughter read The Lord of the Rings, she connected deeply with Samwise Gamgee—not the actor, but the character: his loyalty, humility, and quiet strength. Had she seen the films first, would Sam be anyone other than Sean Astin? Would his character traits be as memorable? Online, many fan-created versions of Hermione, with wild, tangled hair, resemble my daughter more than Emma Watson. Rowling’s canon says Hermione has “brown eyes, frizzy hair, and very clever.” Because so many have read the books, their varied representations don’t conform to what we've seen onscreen. Do we believe Alex Karp, co-founder of Palantir, named the company after ‘Seeing Stones’ because he saw how Peter Jackson visualized them in the films?

AI image generation offers instant novelty: type text and immediately receive a photorealistic oil masterpiece, replicating what your imagination could have done. But pause. Doodle first. Imagine it yourself. Let your own version emerge. It will resonate longer and might even sharpen your prompt.

We Scan More

We are normalizing scanning. Load a long PDF into Acrobat, and its AI offers to summarize it. Paste a blog into an AI tool and it suggests lists, bullets, and headers—less for clarity, more for speed.

But learning isn’t just summarizing. AI can give us a quick “101,” but we risk losing depth. Tools like Perplexity and other AI search tools offer relief from SEO and sponsored slop, but original sources, such as books, papers, and long-form journalism, still matter more than summaries. As films aren’t books, AI isn’t a replacement for authentic human thought.

We Market More

We are choosing to market over inform. Even Harvard Business Review plays this game. A recent LinkedIn post titled “To see the future of Competition, Look at Netflix” teased: “What Netflix understands about business…” It featured the logo and sounded timely. But the article? From 2018. Before Disney+ and AppleTV, before COVID, before AI. The piece was still smart and valuable, but without historical framing, the post misled. A more insightful headline might be: “What Netflix Understood in 2018—And Still Understands Today.” Subtle, but meaningful. It might mean fewer clicks. Clicks are easy. Context takes care.

Image Caption:

A LinkedIn post from Harvard Business Review that features an article from 2018, which is presented as current, relevant knowledge.

We Have Agency

In each case, we’re not just recipients, we’re active participants. We have the power to make AI work for us. It’s not yet engineered for engagement at all costs. Deep thinking doesn’t emerge from short-form videos or SEO, from what we scan or are enticed to click.

When I use Perplexity, I prompt it to include credible, original sources, and I click and read those sources before reading Perplexity’s summary. When writing with AI, I specify no emojis, no em dashes, no bullets. I ask it to improve flow, not override intent. I follow Strunk and White’s advice for writing clarity first, style second, and the primacy of the paragraph.

Language powers both AI and authenticity. They don’t have to conflict. My new book, Practical AI Style: Essential Elements for Developing AI Style, launches June 24. It offers practical methods to develop your AI style, shaping AI with integrity, so we don’t just let it shape us. And if you read this far, it might just be worth reading.

Image courtesy of wikicommons.