Skip to main content


Google AI is a hot sticky mess. Now they're telling people to add glue to pizza, as it took a shitpost from Reddit.

Credit: https://www.threads.net/@petergyang/post/C7S6fzINqZj

in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

ohh no glue onto pizza well I'm trying tell ya just something's in this world just isn't right and I'd bet ya dollar to dog biscuit this is definitely one of those.
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

it's ok, legal is all over this, and they say that just adding a small print disclaimer somewhere about "Google doesn't guarantee the safety or accuracy of these results" is all that's needed
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

A hot sticky mess suggesting how to make another hot sticky mess.

Somehow, that makes perfect sense to me.

in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

AI overviews are disaster because of the low quality content Google Search is shoveling us into.
Gemini 1.5 Pro is good enough for many tasks, if you give it right sources of truth, like pupils books or arXiv articles
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

I'm not 100% sure, but adding glue to make the cheese tackier sounds like one of those tricks food advertisment photographers use to make the food more visually appealing/last longer on set.
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

Ah, you don't know that.

That's the beauty.

In general, without special precautions, you never know where the LLM comes up with the probabilities for the next token.

in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

the quick questions (or how they're called) also spit out more shit than facts, I remember one where they quoted an off-topic forum on chess.com for a someone with 400 IQ
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

soon we will have warning on package like that "DO NOT USE UNDER THE RECOMMENDATION OF AN AI"
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

But, but, but... I was told that "AI" is the future. Its smarts and deep reach of knowledge are gonna save us all, and further the light of human consciousness...

#LLM #DTBO

#LLM #dtbo
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

I mean, it makes as much sense as using autocorrect to generate meaningful answers to questions.
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

with the release of AI we opened a can of worms......no wonder they make movies where the AI takes over the world......hahaha #destiny
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

This is known as Data Poisoning and it's a nice to see it actually works.

There are other ways to buck LLMs too.

Just sayin'.

๐Ÿค™

in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

people that do not understand sarcasm are dangerous; technology that does not get sarcasm is even more dangerous by its mere unlimited
propagation
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

This made me laugh and being 72 years young I remember eating the white glue we used in grade school. It tasted minty fresh!!๐Ÿ˜‚๐Ÿ˜‚๐Ÿ˜‚
in reply to Liam @ GamingOnLinux ๐Ÿง๐ŸŽฎ

I see stuff like this and think of the techbros in the belly of the #LLM like Charlie Allnutt kicking the boiler on the African Queen.
#LLM

Lo, thar be cookies on this site to keep track of your login. By clicking 'okay', you are CONSENTING to this.

โ‡ง