Nick Tune 🇺🇦May 9, 2023
Will Oremus

I went on a quest to find out if AI is already secretly writing a lot of the stuff we read online. I found more than I ever imagined.

From online recipes to mattress reviews to celebrity horoscopes, human writing is fast becoming the exception on the internet, rather than the rule.

My story today: https://www.washingtonpost.com/technology/2023/05/05/ai-spam-websites-books-chatgpt/

He wrote a book on a rare subject. Then a ChatGPT replica appeared on Amazon.

The web has long been full of spammy content. AI is on the verge of making it much worse.

The Washington Post
May 5, 2023 at 1:34pmWeb
Show threadMullaneyMay 5, 2023
@willoremus Excellent story. I hope that search companies are working hard to figure out ways to filter out work that is substantially ai generated. I think there is a huge opportunity for the search engine that does it first and does it best.
Show threadCreesMay 6, 2023
@mullaney @willoremus more than filtering out, I believe the exercise is for the society to accept AI as the tool it is and correctly label their products as “made by blahblahAI”, “with the help of blahblahAI”, etc. Behind the AI there’s still a human responsible for the product ending on the (virtual) shelves.
Show threadLisPiMay 6, 2023
@mullaney @willoremus Cutting the majority out is actually pretty trivial. Downranking sites that rely on a lot of Javascript related to ad-tracking or ads in general would do the trick for basically everything that's not disinformation at publisher's cost.
Show threadi_give_u_wormsMay 5, 2023
@willoremus
"why do all these recipe sites start with an ever expanding set of paragraphs that make no sense"
was google the last to know or did they just act like it??
Show threadSven S. SvenssonMay 5, 2023
@willoremus I’ve gotten that feeling while surfing around as well. I’ll do searches to read about a piece of guitar gear and it feels like a good amount of websites are written by bots. The recipe sites are the same.
Show threadJacob TorreyMay 5, 2023
https://mountaincommunity.co/objects/d51dc685-6e18-4e3c-8ef0-cf0624c26666
Jacob Torrey (@[email protected])

Built a little #browser #extension today that sets the opacity of each paragraph's text to the confidence of a #generative #AI detector model that the paragraph is human-generated. AKA AI Noise-can...

Show threadKaren BethanyMay 5, 2023
@willoremus
I suspect it’s also creating a far larger share of the cheap/free book content getting published on Amazon than folks realize.
Show threadsorrykbMay 5, 2023
@willoremus Which is, not coincidentally, why search returns such awful results now.
The internet is mostly garbage, so search results are mostly garbage.
Show threadsorrykbMay 5, 2023
@willoremus well, partly why. The other reason is the existence of companies like this, who along with AI companies are working diligently to make the internet worse for everyone except skeevy advertisers.
Show threadⒶnarchiaslMay 9, 2023
@sorrykb
Spambots have been like that since the early 90s. Ever since search engines emerged, it has been a race of filtering spam and circumventing the filter while applying SEO.
Now that spambots have a tool to circumvent the human bullshit filter, the floodgates have opened spaces which seemed safe before.
@willoremus
Show threadJill MinorMay 5, 2023
@willoremus @gwendolenau well, good. That, plus the total absence of ads and popups and trackers should make my blog more useful.
Show threadPeter ButlerMay 5, 2023

@willoremus The claim that “everyone is reading AI-written articles” seems bunk, at least if they mean articles *fully* written by AI

Technical writing, for sure. Astrology and celeb bios, for sure. But fiction, poetry and journalism, no, or at least not without *significant* human involvement

There’s no way to tell if an article was “written” with AI and then cleaned up by a human. What if an AI generates the H2s? AI written? Spell checkers? SEO keyword optimizers? We’ve had “AI" for decades

Show threadnick colemanMay 5, 2023
@peterbutler Agreed, I think some of the fears of AI are misplaced but at the same time it is quite eerie…
Show threadTara HMay 6, 2023

@peterbutler @willoremus

I've been using AI to help with writing for a while - not to write things for me but if I'm hitting a brick wall or the like. I'm sure there's folks just using it to churn out posts, but I feel like a lot of what I've seen people talking about using it for has been helping with the writing process when stuck

Show threadPeter ButlerMay 6, 2023

@THLiterary @willoremus It works well for brainstorming, just like wordhippo works as a great thesaurus

It’s also good for writers who cover a particularly narrow beat. They can pump in all their heds and leads and see what they are missing — or what people are searching for — tools like questiondb.io are useful for that too

Articles are definitely getting cranked out completely by AI, as that news investigation showed, but nothing good … that I know of … yet

Show threadMichael SokolovMay 5, 2023
@willoremus well done. Are we in the uncanny valley yet? Hoping a backlash will re-value humans doing good writing
Show threadCathy Caine May 5, 2023
@willoremus So is AI the one with all the errors or the one that is correct?
Show thread防空識別區ᵒᵘᵗᵉʳᵏᵒˢᵐᵒˢMay 5, 2023
@willoremus I'm not going to read nor share your paywalled article.
Show threadPatrick GillamMay 5, 2023
@adiz @willoremus
Did AI write this
Show thread防空識別區ᵒᵘᵗᵉʳᵏᵒˢᵐᵒˢMay 5, 2023
@patrickgillam Yes. ​​ @willoremus
Show threadDantali0n  May 5, 2023
@willoremus just google any "top 10 of x or best of x 2023" literally each of these "review" sites is autogenerated garbage
Show threadTeemuMay 5, 2023

@willoremus hope this will eventually ignite interest on content ”quality” (whatever that means to people).

Or do we end up extending ad blockers to filter out ai-generated content? Who knows.

Show threadLisPiMay 6, 2023
@tfri @willoremus Most of those content farms rely on ads and subscription data selling for profit so...avoiding those two things would do a lot to help.
Show threadMad Dog Ace RunMay 5, 2023
@willoremus
We are the Borg. Lower your shields and surrender your ships. We will add your biological and technological distinctiveness to our own. Your culture will adapt to service us. Resistance is futile.
Show threadRalphieboyMay 5, 2023
the entire nature of school will change: students will receive instruction at home from ChatBots and only attend school to do their homework and take tests in a supervised setting with limited Internet access.
Show threadindyradioMay 5, 2023
@willoremus They haven't completely taken over print yet. Book printed before 2022 will eventually command a premium. The destruction of 100s of thousands of books at UCSC wille eventually be understood as one of the great catastrophes of history.
Show threadDaddyMay 5, 2023
@willoremus I suspect it’s been good enough to write web search optimized articles and product reviews for sometime now.
Show threadShoq 🌿May 5, 2023
@willoremus I had a hunch this article was coming. I have a bad feeling this is all moving way too fast and there's no way to slow it down.
Show threadJW prince of CPHMay 5, 2023

@willoremus 👆 someone finally found some of those successful AI-driven businesses I asked about recently - and, in news that should surprise nobody, they're A. not new, and B. everything that makes the internet suck sweaty donkey testicles...

For marketers shitting their pants: If you weren't doing that before the current spate of chatbots, you've no particular reason to do so now, but you do you.

For everyone else: If you suspect this tech will ruin everything, it already does.

Show threadSheila M. Averbuch - AuthorMay 5, 2023
@willoremus great piece, Will
Show threadJim colahanMay 5, 2023
@willoremus we’re you aware your post article on Ai is only available to WAPO subscribers? Can you cut and paste it on line Mastodon?
Show threadJen Roth ⚾ 👟May 5, 2023
@Jdcolahan Here, have a gift link: https://wapo.st/3ntG1Pd
He wrote a book on a rare subject. Then a ChatGPT replica appeared on Amazon.

The web has long been full of spammy content. AI is on the verge of making it much worse.

The Washington Post
Show threadJim colahanMay 6, 2023
@tealdeer thank you.
Show threadnick colemanMay 5, 2023
@willoremus this is absolutely insane as a writer and honestly scares me. I know AI will never have the power to fully replace the human mind but still, it’s eerie.
Show threadSane ThinkerMay 6, 2023
@willoremus This is the last frontier of hanging onto humanity, the theft of our very thoughts. AI is the coup de gras, the last blow before our death by technology.
Show threadJeff MoeMay 6, 2023

@willoremus

Why don't these articles point out that mining all this content to build models is copyright piracy, with criminal penalties?

So easy to fix: the people using pirated content to generate models can just have the FBI run over their houses with bearcats. Just treat it like they are sharing music.

Show threadOllie FrancisMay 6, 2023

@willoremus "From online recipes to mattress reviews to celebrity horoscopes, human writing is fast becoming the exception on the internet, rather than the rule. "

- That's what will make it special.

Show threadMark-André TremblayMay 6, 2023
@willoremus of course, everything that was within the listicle trend has probably switched to AI writing by now.
Creators of that content always went for the path of least effort.
But to me that's a cause for worry.
It will eventually make real content, like research papers and scientific literature stand out a lot more compared to that echo chamber of "useless" post.
#hope #aiwriting
Show threadKanaMaunaMay 9, 2023
@willoremus This past week I found myself with a bag of fresh turnips. Having no clear idea how to cook them, I decided to look up a nice easy recipe for grilled turnips. What a wild ride that was, recipes with long repetitive introductions followed by contradictory and wrong instructions.
Show threadRafa de CastroMay 9, 2023

@willoremus Fully agree on this. Content creation will have to change. When we have automatically generated blog posts the most evident strategy is just create a shitload content just to see if you are lucky enough something sticks. It's the same strategy than with the email.

Scams will soon follow.

Show threadRaven OnthillMay 9, 2023

@willoremus they're going to get people killed, you know.

If Amazon, Google, and MS loose a ton of money on this, I won't weep.

Show threadDock BunsonMay 9, 2023
@willoremus yep. Bots scrubbing bot-written content for content.

Bots on bots on bots.

The fediverse is the only online space that I feel that I can communicate with random people & believe that they are actually human.
Show threadRichard HulseMay 9, 2023
@willoremus great story Will, and slightly terrifying. I can imagine those content farms that used to have duplicated stories now increasing in number and containing slightly different versions of the same story.
Show threadFelicitas MacgilchristMay 10, 2023
@willoremus @festal Yes to this: “As AI writes more and more of what we read, vast, unvetted pools of online data may not be grounded in reality, warns Margaret Mitchell, chief ethics scientist at the AI start-up Hugging Face. “The main issue is losing track of what truth is,” she said. “Without grounding, the system can make stuff up. And if it’s that same made-up thing all over the world, how do you trace it back to what reality is?””
Show threadFelicitas MacgilchristMay 10, 2023
@willoremus @festal Isn’t it ironic that none of the science fiction with sentient AI has them making up random stuff, or creating spam. Not even the #feminist #sf I love reading.
Show threadfelix stalderMay 11, 2023

@discoursology @willoremus I see this more as a sliding along a scale. A lot of stuff on the web has already had no relation to truth (or an honest attempt to relate to an external reality). A lot of stuff has been shaped by SEO strategies, click-bait, info wars, corporate PR, or was located somehwere between ads and propaganda.

But, of course, taking all of this as "ground truth" makes things worse.