Automated accounts and video scripts increasingly use undisguised AI text.
As more people come to recognize the tells of AI, which mostly happens as you start to work with AI a lot, the scales are going to fall from their eyes and they are going to realize what some of us already see: how much of this site (and blog posts, articles, papers) are AI now.
Users criticize AI content proliferating across websites, feeds, and YouTube as intolerable lazy spam and noise that erodes authenticity, while a few praise specific accounts for remaining data-grounded.
No Digg Deeper questions have been answered for this story yet.
@emollick So many huge accounts on here post fully generated posts and articles now, even ones that have nothing to do with AI. They don't even bother to hide the most basic tells.
As more people come to recognize the tells of AI, which mostly happens as you start to work with AI a lot, the scales are going to fall from their eyes and they are going to realize what some of us already see: how much of this site (and blog posts, articles, papers) are AI now.
It makes many online spaces intolerable. If I want to talk to ChatGPT or Claude, I'll just talk to ChatGPT or Claude, I don't need to talk to ChatGPT and Claude pretending to be DoofWarrior123 on X with mediocre prompts.
@emollick i’ve even noticed it with youtubers. clearly just reading ai scripts.
somehow feels a step worse.
You can use AI to help with writing, but you need to actually do some writing to be helped with. "Write me a post about [topic], make it really good and interesting" is not going to cut it.
@emollick @Orwelian84 This is why memes matter more than ever. Goblins have become part of the AI lore and culture. The first fun and emergent memetic figure that came from the LLLMs.
@iruletheworldmo Yes, I hate that most of all.
@emollick Wait until AI figures out how to spot AI and we’re cooked.
@emollick I don’t agree with this at all: the AI you talk to depends almost exclusively on the context it has. You and I feed our AI pov and context from a very narrow scope. Having AI to compress and communicate content and context I never thought of is enrichment
@emollick That’s been the case for me. You can’t unsee it. But as someone who used em dashes before AI ruined them for me, not all AI tells are accurate.
@emollick The tell I notice first is missing residue: no constraint, no scar tissue, no weird local detail, no price paid for the opinion, just fluent shape with the fingerprints sanded off.
@romainlalanne @emollick it could actually change language with people adopting certain turns of phrase or on purpose misspellings to set them apart
@emollick Doofwarrior123 is objectively a great handle... can't possibly be AI
@0xLCZ @emollick So far it's been awful at it, and even given strict instructions and regex patterns to avoid (eg those relating to "it's not X. It's Y"), it'll spot the patterns AND STILL use them. Can a tiger change its stripes?
@BiblicallyAccAI @emollick Jeses crist who cares about form? I am talking about content. And how utterly myopic will you become when you are your own stylistic echo chamber. At least give it the writing voice of Ursula K LeGuin. Or Anais Nin
@emollick Here's a typical one now replying to your post. It replies with an anodyne statement that sounds vaguely intelligent, but which contributes nothing to the conversation. And its bio confirms.
@emollick what about “.. really *really* good and interesting”?
@Nymne @emollick If you're working with AI that closely, then give it your unique writing voice. This is less than an hour of work, especially with writing samples.
@AndrewCurran_ @emollick Yeap
@theslowtell @emollick That’s why my prompts alway instruct LLMs to intentionally include typos and misspellings. There isn’t a more believable marker of authenticity.
@Michae1Hinrichs @emollick couldn’t agree more. which is why i tried to provide my own treatise on how to think about them and apply them responsibly. unlikely to gain any serious traction, ofc, because it asks that people take a measured and conscientious approach to these things: http://github.com/da5ch0/kybernetes
