ChatGPT wraps up with an avalanche of cliches . . .

Published 2025-02-09

An example of just how bland AI output is compared to a creative human.

Here is a wonderful article capturing why we need humans doing creative work in a world becoming saturated with AI slop. If you care about fostering a world where human creativity is still valued then June Casagrande’s article “A Word, Please: Coffee-shop prompt stirs ChatGPT to brew up bland copy” is well worth a read.

She observes,

ChatGPT bases its writing on lots and lots of online documents and databases, guessing what word should come next based on the words all those other articles used next. The result is a very slick, polished voice, along with a ton of cliches and empty sentences. But ChatGPT’s biggest weakness is that it has no idea what interests humans. What’s more, it doesn’t understand what words mean.

Here is but one example she gives,

Now the AI program’s third sentence: “With a steadfast commitment to quality and community, Caffe Maximo’s expansion into Redondo Beach symbolizes its dedication to fostering local connections.”

In a book I published in 2010, I spent a whole chapter explaining why writers should not do this, which I will summarize here in two syllables: Who cares? Such a vacuous collection of words wastes the reader’s time and, in a tight article, wastes precious space too.

This gets at the heart of the problem with most AI generated content — it takes up space and wastes time on insignificant ideas.

A creative human cares deeply, and this care will manifest in the expression of their art. Generative AI is, at the deepest technical level, simply predicting what is most likely. Truly unique and creative expression is the opposite of likely. It takes a tremendous amount of effort.

The problem with bland AI slop is eloquently summarized by June in one of her final statements, “ChatGPT wraps up with an avalanche of cliches . . .”

3 Other Things

  1. Apple is telling it’s AI to not hallucinate. Makes sense, it allways works when I tell my child to not lie 🙄
  2. Researchers are experimenting with AI systems to do science, revealing connections between data previously unseen. This is really interesting work. However, going with the theme of June’s article above, most of my favorite scientific breakthroughts came from the creative art of experiment design.
  3. A paper claiming AI can be more creative than a human, “The creative ideas produced by AI chatbots are rated more creative than those created by humans.”