The Night Flight @constellationsoffreckles - Tumblr Blog

if you're having trouble sleeping the best you can do is put a bright object close to your face and look at it for at least 30 minutes. if that doesn't work you can close your eyes but make sure to think really hard about a bunch of bullshit

drizzledrawings

☆tiger☆

mushbeast

My Child

odyver

alright I've got to do some quick math to explain attitudes towards AI to my boss.

we're looking to create an AI policy, and when we were talking about this, my boss (older millennial) was genuinely shocked to hear that younger people do not (seem) to view AI positively (a la the recent commencement speakers being booed)

please rb for larger sample size!

Question 1/3

What is your age, and do you feel AI is a net positive or net negative in our lives today?

under 18, AI is a net positive

under 18, AI is a net negative

18-29, AI is a net positive

18-29, AI is a net negative

30-45, AI is a net positive

30-45, AI is a net negative

46-60, AI is a net positive

46-60, AI is a net negative

over 60, AI is a net postive

over 60, AI is a net negative

Voting ended onJun 11

odyver

Question 2/3

How often do you visit or interact with museums/archives (whether in person or online)?

Frequently (multiple times per month)

Often (multiple times per year)

Occasionally (a couple times per year)

Rarely (once every couple of years)

Never :(

Voting ended onJun 11

odyver

Question 3/3

If you saw a museum was using AI in exhibits, marketing, research, etc., would you be more or less inclined to visit that museum?

under 18, more inclined

under 18, less inclined

18-29, more inclined

18-29, less inclined

30-45, more inclined

30-45, less inclined

46-60, more inclined

46-60, less inclined

over 60, more inclined

over 60, less inclined

Voting ended onJun 11

Thank you for helping with this data collection. Please rb for as big a sample as possible!

🫶

rainbowgaez

ok guys but imagine how hard it's gonna hit if something good ever happens again

hollow-head

"who do you self insert as when you read?"

This is me when I read:

campsis

Eden Kalif, Good Cats

jack-of-blades

My housemate's cat came into my room while my dictation was on...

charl0ttan

huge fan of the depth of a good purple but another area that draws me is definitely around aquamarine/turquoise/seafoam. you can not go wrong once the green starts getting just a tinge more blue. a gal could certainly do worse than to pull over there and stay a while

charl0ttan

something earth shattering going on here

emily84

this is why one of my favorite all-time paintings is Ship in Stormy Seas by Ivan Aivazovsky... he was really onto something there

a close up to just... light shining through those waves, makes me feel faint with exhilaration every time

is-there-a-boat

THERE IS A BOAT BY IVAN AIVAZOVSKY!!

clayandsorcery

Ivan Aivazovsky could paint glowing water. One of the GOATs for sure.

loucifersbitch

if you need me, i’ll be sobbing on the floor. humans, man

my-dailydreams

c3rvida3

Me and my cousin have an ongoing bit where we pretend we made "slightly better" versions of things where we'll be like.

"That was a pretty good movie, but not as good as my movie, House of 1001 Corpses," or "I guess this song is okay. Kind of reminds me of a song I'm working on called 'Faster Car'."

Never once has it been funny or made anyone but us laugh.

mucmucvi

This reminds me of a joke I have with two of my cousins

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?