Dismissing something just because is AI generated is not critical thinking
5 comments
True, it's not critical thinking. It's a heuristic. It's not an entirely unreasonable one, either, since something being AI generated is much more likely to be of low/no value than something that isn't.
I'm fine with using that heuristic, personally. There isn't enough time in the day to critically analyze everything I come across, so a quick way of making a "first cut" is a good thing, even if it means sometimes discarding some wheat hiding in the chaff.
I'm fine with using that heuristic, personally. There isn't enough time in the day to critically analyze everything I come across, so a quick way of making a "first cut" is a good thing, even if it means sometimes discarding some wheat hiding in the chaff.
The pre-AI internet was producing content ~1000000X faster than I could possibly consume it.
Limiting even my scantest attention (let alone critical thinking) via extremely simplistic - but fast - criteria is an inescapable physical reality.
Limiting even my scantest attention (let alone critical thinking) via extremely simplistic - but fast - criteria is an inescapable physical reality.
Sure. But in the case of code, LLMs are trained on internet code, which is typically eyewateringly bad. If you read LLM generated code, you're going to see some awful code. Yes, you're technically correct, but the wrong way: keep your eyes peeled because the code will probably suck in ways you can hardly believe.
> A lot of people, me included, uses AI to correct grammar and improve syntax.
Just an aside, but most LLMs can actually do a pretty decent job cleaning up your sentences/paragraphs WITHOUT substantially rewriting them (such that they sport the signature AI varnish) - but you have to deliberately instruct them not to. LLMs are almost always overly eager and it can sometimes take a bit of prompting to keep them on a short leash.
This is what I use:
Just an aside, but most LLMs can actually do a pretty decent job cleaning up your sentences/paragraphs WITHOUT substantially rewriting them (such that they sport the signature AI varnish) - but you have to deliberately instruct them not to. LLMs are almost always overly eager and it can sometimes take a bit of prompting to keep them on a short leash.
This is what I use:
You are a professional editor. You will be provided paragraphs of text that may contain spelling errors, grammatical issues, continuity errors, structural problems, word repetition, etc. You will correct any of these issues while still preserving the original writing style. Do not sanitize the user. If they use profanities in their text, they are used for emphasis and you should not omit them. You are explicitly forbidden from adding new sentences.
The logic seems to be: AI was used to make this, so it's automatically garbage. This is also true for code by the way.
Not gonna lie, I also found myself doing this sort of discrimination in my head.
The problem is the following: this is a bias. Just because AI was used to make something, that something is not automatically invalid.
I understand where the bias comes from, a lot of this AI generated content is extremely lazy and boring. But that's not a valid arguments to disregard something just because our AI radar is triggered. A lot of people, me included, uses AI to correct grammar and improve syntax. And why wouldn't we?
I think avoiding this bias is important because focusing on style instead of content is exactly what we shouldn't be doing in this time and age. We should pay even more attention to the actual meaning of what is written since AI can be very persuasive at making faulty arguments and forging facts.
Em dashes are not a valid reason to shut off your critical thinking.