The SLOP Problem: Will LLMs Replace Thinking?
As a writer, I’ve been reflecting on the usefulness and challenges of LLMs and reading various perspectives to understand how others view this. One of the main advantages of LLMs is how they make “creating” text easy. However, this same ease is frequently misused to mass-produce repetitive content that adds little to no value to conversations or discussions (referred to as AI SLOP). When people publish such output, it drags down the overall quality of writing in the content space.
This mirrors what happened with the advent of email. Email transformed direct communication, but it also enabled spammers to flood inboxes with worthless messages that wasted time. Like spam in email, the misuse of LLMs risks flooding the digital space with low-value content, making it harder to find thoughtful, meaningful contributions. Just as spam filters became necessary to manage email overload, will we see new tools to identify and filter out low-quality, AI-generated content from personal digital spaces?
That said, I have found some valuable use of LLMs as a writer:
Good Uses:
- To process some of my thoughts: I use LLMs to ask challenging questions, explore alternative reasoning, and find original references that I can verify independently. I do recognize that it is NOT an alternative to having real conversations and discussions with people with experience.
- To refine text that I wrote: I sometimes use LLMs to generate multiple variations of a text I’ve written. Sometimes, it gives me a better
Harmful Uses:
- Passing off AI-generated text as one’s own: Simply to generate text that you pass on as being the “author” — essentially cluttering the digital space with low quality, repetitive text and potentially even engaging in plagiarism. As Mark Dingemanse wrote — “Writing is thinking” and “to outsource writing is to give up on thinking”
For further reflection, I highly recommend these insightful writings by Mark Dingemanse:
The key isn’t whether LLMs are good or bad, but how we use them. At the end of the day, LLMs are tools that can be used either thoughtfully or carelessly. When people use LLMs to skip the thinking process and simply publish AI-generated text as their own, they’re not just potentially plagiarizing — they’re avoiding the crucial connection between writing and thinking which reduces quality.