LLMs aren’t writing
I’m not an AI hater: I’ve used it for my fair share of projects that involve coding or for troubleshooting various technology problems. In my work testing Apple’s Apple Intelligence releases, I’ve generated some images, all truly bizarre and many of them good for nothing beyond a joke.…
The author expresses surprise at the strong reactions to Anthropic's AI watermarking scheme, particularly after reading a passionate rebuttal from friend John Gruber. Gruber derides the watermarking system as "a perversion of writing," but the author argues that large language models (LLMs) are far removed from the meticulous and intentional nature of human writing.
LLMs do not possess the ability to care about the words they select, as they lack any form of awareness or understanding. The author emphasizes that LLMs generate text through statistical modeling, without any real comprehension of meaning or intent.
The author uses the example of word choice, such as "leaped" vs. "jumped," to illustrate the nuanced differences in meaning that LLMs cannot grasp. Since LLMs lack a predefined "correct" answer, their choice of words is purely probabilistic, lacking any inherent correctness. Similarly, the author questions the relevance of a watermarked substitution like "pineapple" for "banana" when the original context requires a specific fruit.
The author acknowledges that AI-assisted tools like autocorrect and autocomplete have become commonplace, but argues that this is simply a progression of technology's growing influence on our lives. The author contends that frustration with LLMs' word choices is misplaced, likening it to anger over a larger, more destructive force like Godzilla. The author emphasizes the far greater societal, environmental, and ethical implications of these broader technological advancements.
The author concludes by criticizing the equating of AI-generated text with human writing, arguing that comparing the two is a flawed analogy. While AI may consume and regurgitate existing text, it lacks the creativity, intentionality, and emotional investment that characterize human writing. The author suggests that if people are upset about AI-generated text, it may be a sign that they are comfortable with its output, and potentially less motivated to produce their own high-quality writing.
Written by urgent.news from Six Colors's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
