i like my coffee

dark, bitter and hot | multilingual multifandom shitposting | suomitumppu | occasional artist | melodramatic millennial | goth | kaaoksen palvelija | ace gay | any pronouns

[pupucino]

muistutan vaa, mul ei sit oo tumppuu

[tothechaos]

hate when people say something along the lines of "we must get along so well because you are/i am [some astrological sign]" to me. what are you fucking talking about. we get along because im fucking nice. a decision which i regret when you say shit like that

[alaitkeiltoja]

“About kolkyt sekuntia ja ei oo punaviinist jälkeekään!”

—

minä, samastun pesuainemainokseen (via syksyksi)

@tragid

(via henkari)

[surkeart]

comic about peas🫛

[klfg]

[source]

[klfg]

🚨BREAKING: OpenAI published a paper proving that ChatGPT will always make things up.

Not sometimes. Not until the next update. Always. They proved it with math.

Even with perfect training data and unlimited computing power, AI models will still confidently tell you things that are completely false. This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

And their own numbers are brutal. OpenAI's o1 reasoning model hallucinates 16% of the time. Their newer o3 model? 33%. Their newest o4-mini? 48%. Nearly half of what their most recent model tells you could be fabricated. The "smarter" models are actually getting worse at telling the truth.

Here's why it can't be fixed. Language models work by predicting the next word based on probability. When they hit something uncertain, they don't pause. They don't flag it. They guess. And they guess with complete confidence, because that's exactly what they were trained to do.

The researchers looked at the 10 biggest AI benchmarks used to measure how good these models are. 9 out of 10 give the same score for saying "I don't know" as for giving a completely wrong answer: zero points. The entire testing system literally punishes honesty and rewards guessing.

So the AI learned the optimal strategy: always guess. Never admit uncertainty. Sound confident even when you're making it up.

OpenAI's proposed fix? Have ChatGPT say "I don't know" when it's unsure. Their own math shows this would mean roughly 30% of your questions get no answer. Imagine asking ChatGPT something three times out of ten and getting "I'm not confident enough to respond." Users would leave overnight. So the fix exists, but it would kill the product.

This isn't just OpenAI's problem. DeepMind and Tsinghua University independently reached the same conclusion. Three of the world's top AI labs, working separately, all agree: this is permanent.

Every time ChatGPT gives you an answer, ask yourself: is this real, or is it just a confident guess?

[rederiswrites]

An article confirming and elaborating on the above.

[helenvaughans]

one of the really vile things about generative ai which i feel like we don’t talk about enough is that for like 90% of the cases where it’s used to generate text, image, video etc. THE EXPLICIT INTENT is to deceive the viewer/reader into thinking it’s NOT ai. what this does is to create an extra level of ambient distrust when viewing any kind of media not created by yourself and when you do notice that something is ai-generated, there is the implicit understanding that whoever “made” the ai-generated thing was trying to deceive you or take advantage of you. thus a kind of ambient paranoia is spread across society where everyone feels like everyone else is out to get them and that nothing is real.

[quasi-normalcy]

I actually think that science fiction has done everyone a disservice by presenting escaping to another planet as a remotely feasible near-term solution to problems on Earth.

[quasi-normalcy]

Like, I’m sorry, but if terraforming Mars is an option then why can’t we just terraform the Earth?

[gremlininho]

[beepbeep-eddy-27]

[swagdinner]

[runcibility]

[pupucino]

mun saatetaan kuulla sanovan et parast kesäs on keskikesän yöt, mut joka kerta ku heinäsirkat ja hepokatit alkaa soittaa, mietin vähä et tää on kyl parast kesäs 🦗

[mademoiseli]

The notes are broken. This is what tumblr is all about apparently.

[atonau]

THE NOTES ARE BROKEN!  This has been reblogged so many times, Tumblr just shrugged and said “infinity”

[sonandheirofnothinginparticular]

[leyilay]

Johnny at the café

有参考

‖LeyiLAY.2025.4

[commander-ledi]

fazerin suklaalevy jossa on buranaa täytteenä

[kinuskikakku]

pirkan suklaalevy jossa on ibumaxia täytteenä

[wizardshark]

harry potter fans in 2025

[battlecrazed-axe-mage]

We've all gotten just a bit too comfortable being jerks to strangers on the internet I think

[battlecrazed-axe-mage]

So I've hidden this reply, both because it's obnoxious and because I don't want the person who wrote it being harassed for it, but I need you to understand: I don't know you. We are not friends. This is not fun or cute, we are not sharing a charming joke together. You are just being an asshole.

[battlecrazed-axe-mage]

literally that is what the post is about, I am saying people should be less eager to jump on any chance to be snarky and rude to total strangers on the internet

[sidedere]

DID THE JOURNAL FACTORY FUCKING EXPLODE???

you said it yourself: you're looking to vent it LITERALLY ANYWHERE

so vent it somewhere private. or at least not literally aimed AT another person, a total stranger at that

Like, this reblogger sounds so insanely self centered in their reblog. notice how both options focus on how being rude would affect THEM. "B has no consequences for me so it's perfectly fine to do"

(the only reason I didn't show their username in the screenshot is because, given how self victimizing they sound in their reblog, I believe that, if I did show their username, suddenly online stuff wouldn't seem so inconsequential to them and they'd accuse me of sending harrassment their way and putting them in danger)

[battlecrazed-axe-mage]

You said it better than I could. Of all the inane and ridiculous things I've seen in my notes because of this post, "I NEED to say fuck you to strangers or I will literally die" is certainly one of them