Doremus Schafer: This is actually one of the _less_ outrageous (though still egregious) AI fails, because it's actually just a direct scraping of a human-generated (but satirical) post made four years ago: www.quora.com/What-fruit-n...

Love what google is up to these days

This is actually one of the _less_ outrageous (though still egregious) AI fails, because it's actually just a direct scraping of a human-generated (but satirical) post made four years ago: www.quora.com/What-fruit-n...

Not sure that makes it less outrageous. Actually, not just putting that post there makes it more outrageous bc the AI is both very stupid and a plagiarist

yep. hard proof that it's not "synthesizing" or "analyzing" shit.

They keep telling me that it takes in lots and lots of information and spits out an amalgamation and that is different than plagiarism.

Amalgum

Plagiarisum

Hotshitum

Not just plagiarizing, but plagiarizing from Quora, the site most people want to exclude from their search results

Rip Yahoo Answers

No but seriously even your average Facebook brainrotted uncle is more reliable than Quora

Where you been

It shows the source for the answer prominently underneath this result, but that doesn't make a funny post on social media.

I thought the "and coconut" was too perfect. The setup is just about plausibly AI because it's the kind of little kid joke that relies on slapping together nonsense similar to things they've heard, but the punchline needs understanding of subversion.

whoa, good find! explains the "and coconut" which otherwise would be quite the AI hallucination

I'm off Twitter, but I thought about starting a new account where I would post nonsense to "train" the AI.

That's what most accounts on Twitter are already doing, purposefully or not

Post nonsense in alt text, along with an accurate description.

If I go to google the color of bananas and I get the AI telling me bananas turn blue when they’re feeling shy and red when they’re happy I will blame both of you personally

It could be worth it.

And now Quora is using AI, so this is all turning into the Inhuman Centipede.

Except, bsky.app/profile/nano... So it keeps happening in different ways.

Bing

Something else to think about: AI's plagiarism is discreet enough to turn a numbered list into a sentence one, but not switch up the order? I mean thanks to that the punchline is kept, but this is such a basic plagiarism fail.

That makes it worse! The LLM is taking satire/jokes/etc from around the Internet and feeding it to people without citing a source!

So it’s THEIR fault!

It’s totally fine, because this is the one case in history of the internet where someone was either joking or wrong

Should have known. "Coconut" was too funny to be an LLM.

Writing "cum" in that quora answer and making some popcorn

Bonitum Applum— you got it going on... #Schmience

Gemini has a good result, I wonder why Google doesn't use that

Probably because that's also wrong?

Yeah it ignoring plum is very funny. But I assume the “meme” result comes from people searching it in mass after me out of curiosity But mine was a genuine search 🤷🏻‍♂️ I search for words all the time and before the google LLM I always got normal results

Dim sum, aprium, capsicum if you’re Australian, maybe even gum

Sorghum

Okay where were you doing my witty word play game?! I’m calling dibs on you being on my team from now on