Great Writers Aren’t Obsolete. AI Makes Them More Valuable Than Ever.

Whether they intend to or not, AI companies are eliminating unique answers, because they are seeking to install a reward mechanism that encourages the models to play it safe.

Originally published at American Greatness:

The more I learn about how AI works, the more I’ve come to prize the value of human creativity in the AI era, especially when it comes to writing.

AI is a revolutionary productivity tool, but it is not a replacement for human knowledge or input. The software can only go as far as the human behind the machine.

AI has many incredible applications, but AI writing is mediocre by design. It’s fine. Not great. Just fine.

It’s suitable for the vast majority of the population, especially since most people currently can’t identify the difference between AI writing, AI-assisted writing, and purely human, unassisted writing. Undoubtedly, there’s a place for all of these calibrations.

But there are a few reasons why AI models can’t, on their own, produce uniquely outstanding literature.

Let me explain:

Think of your favorite chatbot as a student who has read every book in circulation. That’s incredibly impressive. Your student is an encyclopedia of knowledge! You ask him a question, and his default job is simply to guess the next word. He’s very good at guessing, but that’s because, despite his wealth of knowledge, he’s inclined to be probabilistic and not unique. Sad! He knows that among a million writers, there is a significant chunk that will finish a sentence the same way, and that’s the decision he will use in his writing. So by default, he is brilliant but seeks to optimize an average. And he can’t, by design, break out of this mediocrity shell.

Major AI companies, like Anthropic, SpaceXAI, OpenAI, and the like, have hired thousands of people to train AI models and evaluate their answers. Whether they intend to or not, they are eliminating unique answers because they are seeking to install a reward mechanism that encourages the models to play it safe. AI is trained to be a teacher’s pet, one who calculates what the “teacher” (AI model trainer) prefers and decides to write that way to achieve the reward. AI writing is habitually mediocre because of this.

Moreover, all of the major American AI companies (and foreign ones, because they distill information from American labs) train their models using the same internet with similar processes. And when smaller upstarts build their own models, they train them on behemoths like ChatGPT, Claude, and Grok, which is much more cost-efficient, but it leads to the same problem. So their habits become everyone else’s habits. Despite the mess they’ve created, none of the big dogs in AI seem to want to hire great writers and editors to fix that institutional headache, probably because there isn’t much ROI involved. You can’t tell chatbots apart by their prose because it’s one big circular learning environment.

When a human being writes an article without assistance from AI, they inject creativity from their own life experience or imagination into the piece. If you tell an AI to “write an article” about something, it does not have an independent perspective, so it fills the space with pieces of information that have been spliced together from a variety of places. Despite the claims of delusional tech bros on the West Coast, AI models do not have the ability to think independently. AI models are very advanced software projects that operate under a very impressive but very non-human paradigm.

Fixing this problem is mostly above my pay grade. More and more content online is now written entirely by AI, so AI models are now training on AI writing at an increasing rate, which likely means that this mediocre writing problem will get worse before innovation arrives to make it better.

Due to the aforementioned institutional hurdles to creativity, it’s become clear to me that great human writing (with or without AI assistance) is needed now more than ever.