Andromeda's Avatar

Andromeda

@andromedaparadox

Witch | Hacker

101
Followers
236
Following
4
Posts
05.07.2023
Joined
Posts Following

Latest posts by Andromeda @andromedaparadox

10.03.2026 18:17 πŸ‘ 13657 πŸ” 3975 πŸ’¬ 31 πŸ“Œ 6
Two witches close to each lit with red light. They are holding hands.

Two witches close to each lit with red light. They are holding hands.

Thicker Than Water. πŸ’

20.02.2026 18:54 πŸ‘ 7922 πŸ” 1310 πŸ’¬ 32 πŸ“Œ 6

they stole an hour from women

08.03.2026 18:11 πŸ‘ 5004 πŸ” 1695 πŸ’¬ 4 πŸ“Œ 14

I will not vote for Newsom because he does not believe I deserve human rights.

If it comes down to JD Vance vs Newsom, who are both rabidly transphobic & support ICE, I will vote for a third party candidate.

Luckily Democrats have another 2 years to make sure we select a better, viable candidate!

07.03.2026 21:15 πŸ‘ 561 πŸ” 90 πŸ’¬ 15 πŸ“Œ 18
Nav Toor
@heynavtoor
🚨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.

Nav Toor @heynavtoor 🚨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.

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?

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?

Post image

⚠️Told you so moment:

OpenAI published a paper proving that ChatGPT will always make things up [...] Always.

They proved it with math.

[...] This isn't a bug they're working on. It's baked into how these systems work at a fundamental level.

More: x.com/heynavtoor/s...

07.03.2026 12:53 πŸ‘ 160 πŸ” 80 πŸ’¬ 4 πŸ“Œ 2
Post image

Cooling Down

08.03.2026 00:20 πŸ‘ 5445 πŸ” 1078 πŸ’¬ 24 πŸ“Œ 1
1
Inquiry for New York Times

Good afternoon,
My name is [REDACTED]
and l am a reporter for the
New York Times. I am researching how much money the Palestinian diaspora community has sent to support Gazan families since the war in Gaza began. Does your team track this figure? I really appreciate it.
With gratitude,
[REDACTED]

1 Inquiry for New York Times Good afternoon, My name is [REDACTED] and l am a reporter for the New York Times. I am researching how much money the Palestinian diaspora community has sent to support Gazan families since the war in Gaza began. Does your team track this figure? I really appreciate it. With gratitude, [REDACTED]


Palestine Diaspora Movement 

Hello [REDACTED]
We do not collaborate with The New York Times.
Your email asks about money sent to Gazan families "since the war in Gaza began." Let us first correct the premise. This is not a war. It is a genocide. Calling it a war is part of the problem. It obscures the reality of a colonized people being massacred by a military machine backed by the United States and defended, sanitized, and rhetorically softened by institutions like yours.
The New York Times has spent far too long parroting Zionist propaganda and laundering Israeli state narratives while Palestinians bury their dead. That kind of coverage does not merely misinform. It helps create the political climate in which Palestinian slaughter becomes explainable, defensible, and forgettable.
So no, we will not be sharing any figures, internal data, or community information with a paper that has so thoroughly damaged its credibility on Palestine. Quite frankly, we would rather chew jeans than help The New York Times extract from our community while continuing to fail it.
Do not contact us again for this purpose, The Palestine Diaspora Movement Team

Palestine Diaspora Movement Hello [REDACTED] We do not collaborate with The New York Times. Your email asks about money sent to Gazan families "since the war in Gaza began." Let us first correct the premise. This is not a war. It is a genocide. Calling it a war is part of the problem. It obscures the reality of a colonized people being massacred by a military machine backed by the United States and defended, sanitized, and rhetorically softened by institutions like yours. The New York Times has spent far too long parroting Zionist propaganda and laundering Israeli state narratives while Palestinians bury their dead. That kind of coverage does not merely misinform. It helps create the political climate in which Palestinian slaughter becomes explainable, defensible, and forgettable. So no, we will not be sharing any figures, internal data, or community information with a paper that has so thoroughly damaged its credibility on Palestine. Quite frankly, we would rather chew jeans than help The New York Times extract from our community while continuing to fail it. Do not contact us again for this purpose, The Palestine Diaspora Movement Team

β€œQuite frankly, we would rather chew jeans than help The New York Times extract from our community while continuing to fail it”

lmaooo GET EM

04.03.2026 15:46 πŸ‘ 2543 πŸ” 729 πŸ’¬ 24 πŸ“Œ 23

It’s an ICE staging center that told its queer and Black customers to go fuck themselves.

03.03.2026 16:07 πŸ‘ 873 πŸ” 243 πŸ’¬ 8 πŸ“Œ 4

There you go. An overt technofascist Trump ally will now own CBS, CNN, HBO, Paramount, Warner Brothers, Discovery, and a big chunk of TikTok

26.02.2026 23:01 πŸ‘ 2326 πŸ” 1074 πŸ’¬ 118 πŸ“Œ 80

To any Democrat who ever said about any issue, "Maybe if we agree with Republicans just this far, they'll stop there," remember that you were wrong, it was obvious at the time, and you really should never believe that again about any issue, ever.

26.02.2026 03:15 πŸ‘ 5052 πŸ” 1696 πŸ’¬ 68 πŸ“Œ 37
24.02.2026 22:02 πŸ‘ 1170 πŸ” 281 πŸ’¬ 15 πŸ“Œ 7
Why do I have to pretend that I'm going to print something in order to save it as a PDF. Why do I have to engage in a little ruse.

Why do I have to pretend that I'm going to print something in order to save it as a PDF. Why do I have to engage in a little ruse.

OP has quotes turned off, but if y'all know me, you know that I saw this and a) immediately knew the answer, and b) am incapable of stating the answer without using the word fuck multiple times.

So, to clarify:
The reason is *fucking Adobe*. That's the reason.

24.02.2026 02:16 πŸ‘ 66 πŸ” 19 πŸ’¬ 4 πŸ“Œ 1
Video thumbnail

This guy gets it.

23.02.2026 20:42 πŸ‘ 23666 πŸ” 5747 πŸ’¬ 453 πŸ“Œ 805
In her presentation, Kilgannon mapped out three non-negotiables in the fight against the so-called gender identity agenda, a conspiracy theory touted by anti-LGBT groups that disavows sexual orientation and gender identity. The first is to β€œdivide and conquer. For all its recent success, the LGBT alliance is actually fragile and the trans activists need the gay rights movement to help legitimize them.” In other words, separate trans activists from the gay rights movement, and their agenda becomes much easier to oppose. As Kilgannon explained, β€œTrans and gender identity are a tough sell, so focus on gender identity to divide and conquer.” For many, β€œgender identity on its own is just a bridge too far. If we separate the T from the alphabet soup we’ll have more success.”

Kilgannon identified a wide coalition of potential allies outside the Christian Right who could confront trans friendly measures. Here’s her advice on how to draw them in:

    Explain that gender identity rights only come at the expense of others: women, sexual assault survivors, female athletes forced to compete against men and boys, ethnic minorities who culturally value modesty, economically challenged children who face many barriers to educational success and don’t need another level of chaos in their lives, children with anxiety disorders and the list goes on and on and on.

In her presentation, Kilgannon mapped out three non-negotiables in the fight against the so-called gender identity agenda, a conspiracy theory touted by anti-LGBT groups that disavows sexual orientation and gender identity. The first is to β€œdivide and conquer. For all its recent success, the LGBT alliance is actually fragile and the trans activists need the gay rights movement to help legitimize them.” In other words, separate trans activists from the gay rights movement, and their agenda becomes much easier to oppose. As Kilgannon explained, β€œTrans and gender identity are a tough sell, so focus on gender identity to divide and conquer.” For many, β€œgender identity on its own is just a bridge too far. If we separate the T from the alphabet soup we’ll have more success.” Kilgannon identified a wide coalition of potential allies outside the Christian Right who could confront trans friendly measures. Here’s her advice on how to draw them in: Explain that gender identity rights only come at the expense of others: women, sexual assault survivors, female athletes forced to compete against men and boys, ethnic minorities who culturally value modesty, economically challenged children who face many barriers to educational success and don’t need another level of chaos in their lives, children with anxiety disorders and the list goes on and on and on.

LITERALLY EVERY SINGLE MOVE IN THE ANTI-TRANS GAME IS DESCRIBED IN DETAIL AT THIS VERY PUBLIC CONVENTION THAT THEY HELD

And there are cis people out there who still think the anti-trans movement was somehow a natural, grassroots expression of legitimate issues.

20.02.2026 18:17 πŸ‘ 1216 πŸ” 418 πŸ’¬ 7 πŸ“Œ 8
Preview
Hackers Expose Age-Verification Software Powering Surveillance Web Three hacktivists tried to find a workaround to Discord’s age-verification software. Instead, they found its frontend exposed to the open internet.

Oops. Turns out Persona, Discord’s age-verification service, was secretly screening selfies against government watchlists AND accessible by the feds. It's backed by Peter Thiel.

β€œThe state wants to see everything. The corporations want to see everything. And they've learned to work together.”

21.02.2026 22:57 πŸ‘ 6088 πŸ” 3888 πŸ’¬ 70 πŸ“Œ 270
Post image

I always thought the whole "they want you dead" thing was just hyperbole. But I guess this makes the entire raw milk, no vax, more asbestos stuff make a lot more sense.

20.02.2026 18:42 πŸ‘ 3600 πŸ” 1136 πŸ’¬ 85 πŸ“Œ 158
Post image Post image

I was a corpse at #katsucon2026
#bloodborne #ladymaria

17.02.2026 05:56 πŸ‘ 187 πŸ” 39 πŸ’¬ 2 πŸ“Œ 1
Post image

πŸ‘€The federal judges in West Virginia, where ICE has been pulling over and detaining people on the freeway, are literally screaming about ICE's tactics.

"It is an assault on the constitutional order," Judge Goodwin says. storage.courtlistener.com/recap/gov.us...

21.02.2026 00:27 πŸ‘ 20341 πŸ” 7491 πŸ’¬ 451 πŸ“Œ 380

fashion is made up and if you want to dress like a nun or a viking that you should do that. if you want to dress like cyberpunk 2077 you should do that. i want to see a nun, a viking, and a cyberpunk walk into a coffee shop and all get in line like it's normal. put the cat ears on shinji

19.02.2026 00:35 πŸ‘ 53 πŸ” 16 πŸ’¬ 2 πŸ“Œ 0

i think lots of people underestimate how badly these β€œAI” companies want people to be unable to afford computers. they view personal computing as a roadblock to their desire for complete end-to-end centralization. the fact that their tech is putting hardware out of reach is not just a happy accident

17.02.2026 17:49 πŸ‘ 4320 πŸ” 2163 πŸ’¬ 2 πŸ“Œ 77

This is far more than a claim. These are facts.

17.02.2026 23:52 πŸ‘ 594 πŸ” 220 πŸ’¬ 6 πŸ“Œ 1

Everybody’s talking about the destruction of consumer electronics and that’s bad, but basically everything in modern life has a computer in it. Hospitals, airports, transportation, shipping, medical devices, farm equipment, and point of sale terminals had a good run

18.02.2026 00:21 πŸ‘ 772 πŸ” 272 πŸ’¬ 22 πŸ“Œ 13

in 2023 when I was organizing parents against our archbishop’s transphobic new school guidelines, one dad asked how loud I thought it was prudent for us to be, given how punitive the Archbishop is. I was like β€œloud enough that all the trans/nb kids at your parish see it” and he was like β€œβ€¦. oh”

17.02.2026 04:10 πŸ‘ 1415 πŸ” 309 πŸ’¬ 1 πŸ“Œ 0

β€œDHS is shutting down because it refuses to obey the law” is excellent framing

13.02.2026 21:13 πŸ‘ 3646 πŸ” 926 πŸ’¬ 20 πŸ“Œ 10
Preview
Seven Billion Reasons for Facebook to Abandon its Face Recognition Plans Meta’s analysis that it can avoid scrutiny by releasing a privacy invasive product during a time of political crisis is craven and morally bankrupt. It is also dead wrong.

Meta thinks now is a great time to launch facial recognition surveillance tech in their creepy glasses because EFF will be too distracted by fascism to notice.

We noticed.

www.eff.org/deeplinks/20...

13.02.2026 22:21 πŸ‘ 1807 πŸ” 691 πŸ’¬ 30 πŸ“Œ 30
Preview
Vim Grammar | Learn Vim Vim Grammar

One reason you should be using the vim text editor is that, other than it making you look at least 34% cooler and sexier, is you have to learn the language of commands it uses, in order to use it.

You have to actively, deliberately, learn just how to close it.

learnvim.irian.to/basics/vim_g...

11.02.2026 22:53 πŸ‘ 195 πŸ” 27 πŸ’¬ 14 πŸ“Œ 22

Revolutionary optimism is the disciplined confidence that because trans lives are forced to invent ways of surviving capitalism, we also possess the skills needed to abolish it.

Despair only serves existing power structures.

Collective struggle for total intersectional liberation now. ⚧️

17.12.2025 21:15 πŸ‘ 630 πŸ” 221 πŸ’¬ 2 πŸ“Œ 12

GM: Charisma check.

Mamdani: [rolls natural 20]

GM: that’s a d6 how did you

Mamdani: [direct to camera] Did you know you can check out board games at your local public library? 😊

07.02.2026 05:01 πŸ‘ 33617 πŸ” 6994 πŸ’¬ 221 πŸ“Œ 182
Letter from Senator Ron Wyden:

Dear CIA Director Ratcliffe,
I write to alert you to a classified letter I sent you earlier today in which I express deep concerns about CIA activities.
Thank you for your attention to this important matter.
Sincerely,
Ron Wyden
U.S. Senator

Letter from Senator Ron Wyden: Dear CIA Director Ratcliffe, I write to alert you to a classified letter I sent you earlier today in which I express deep concerns about CIA activities. Thank you for your attention to this important matter. Sincerely, Ron Wyden U.S. Senator

I don’t like this

04.02.2026 19:49 πŸ‘ 7397 πŸ” 1582 πŸ’¬ 194 πŸ“Œ 471