ardentEscalator

Midway through talking about my symptoms my voice audibly started shaking as I was describing them, clearly upset.

In the middle of my monologue he turns the tablet to face me, closes whatever program he has open and the wallpaper is this fucking collage of pictures of lord farquaad from shrek, lovingly decorated. Dude just sat there placidly smiling at me until I noticed and stopped dead in the middle of a sentence. We sat there in silence like this for like a solid minute before I started wheezing laughing. Before I could even say anything else or process it he picked up the tablet and wordlessly left the room, and I just sat there dumbfounded until the doctor showed up. 10/10 doctor experience tbh

I didn’t own a cell phone at the time to get a photo so this rendition from memory is all I can provide you

no discussion

explanations under cut

but like…Becky, of course LGBTQ+ people existed before Stonewall. We’ve all existed since the beginning of time. But the movement got a shock to its senses, a jump-start, a rocket-into-space when that glass shattered via Marsha P. Johnson, and when Sylvia Rivera was up on-stage protesting guess who was on the sidelines heckling her?

Without a trans black woman, LGBT+ rights would not exist. Never forget. Never “pay it no mind”.

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.

Sorry, I messed up