Gilamasan's Avatar

Gilamasan

@gilamasan

Online artist personified as a cat. He/him, neuro-divergent, Digital and traditional artist. My Carrd for my other socials: https://gilamasan.carrd.co/

779
Followers
3,333
Following
2,549
Posts
12.08.2023
Joined
Posts Following

Latest posts by Gilamasan @gilamasan

Spaces for kids have been destroyed or re-worked into stores, adult spaces aren't allowed to be actually "adult" lest we anger the CC companies, and supposedly neutral spaces are full of gambling/porn ads, disinformation, and right-wing pipelines. An internet that is increasingly not for anyone.

11.03.2026 11:25 πŸ‘ 443 πŸ” 187 πŸ’¬ 3 πŸ“Œ 1

A reminder to call your reps today! KOSA has been sandwiched into the KIDS Act and is still a huge censorship threat. You can use this tool- it takes just a few minutes:

www.badinternetbills.com

10.03.2026 17:30 πŸ‘ 512 πŸ” 681 πŸ’¬ 1 πŸ“Œ 10

A good way to get popular on mid-late 2010s Youtube was to deliberately look up fetish content and then record yourself getting mad that you found fetish content.

10.03.2026 14:05 πŸ‘ 4 πŸ” 2 πŸ’¬ 0 πŸ“Œ 0

Lawsuit any% speedrun

10.03.2026 14:03 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

IMO, this is the best Rayman has ever looked. He's fantastically animated here.

10.03.2026 13:50 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

That's just a giant Gonk Droid.

10.03.2026 13:45 πŸ‘ 0 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

"Zohran Mamdani’s leftist allies malign homeownership"

No they don't. You just made that up.

10.03.2026 02:28 πŸ‘ 4 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

NO NO NO YOU DON’T

GUYS LET ME PLEASE SPREAD THE GOSPEL OF FREE TAX USA

federal is free, state is $16, handles even my chaotic freelancer taxes just fine, same step-by-step β€œdesigned for normies” kind of interface as TurboTax but NOT EVIL

tell everyone you know

www.freetaxusa.com

09.03.2026 01:26 πŸ‘ 12457 πŸ” 7547 πŸ’¬ 13 πŸ“Œ 3

The BBC will do anything as long as it doesn't involve self improvement.

08.03.2026 23:20 πŸ‘ 3 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

adding alt text.

stand up for us before it’s too late, cis folks.

07.03.2026 19:08 πŸ‘ 963 πŸ” 951 πŸ’¬ 2 πŸ“Œ 4

taps the sign of something i said regarding another pixar project

08.03.2026 00:19 πŸ‘ 823 πŸ” 267 πŸ’¬ 10 πŸ“Œ 1

It takes less than five seconds to explain what being gay means. People should stop pretending it's a "difficult conversation"

08.03.2026 01:03 πŸ‘ 2 πŸ” 0 πŸ’¬ 1 πŸ“Œ 0

All these who voted "yes" you have a chance to either contact them to try to convince them that these laws don't work or also vote against them so they don't get re elected

The fight is so neck and neck that a few reps could make a difference

07.03.2026 16:29 πŸ‘ 567 πŸ” 653 πŸ’¬ 1 πŸ“Œ 1
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

There are more steps before this becomes law. It has to go through the committee of rules, then a house floor vote, then the senate. And that takes time. Do not stop calling. Do not doompost. And please plan to vote these asshats out in November

www.badinternetbills.com

www.vote.org

05.03.2026 20:57 πŸ‘ 82 πŸ” 110 πŸ’¬ 0 πŸ“Œ 0
Video thumbnail

We thank @aoc.bsky.social for speaking the truth about age verification in the House E&C hearing today! As she stated plainly, and we agree, age verification is a smoke-screen for censorship. It is NOT about kids safety!

05.03.2026 21:50 πŸ‘ 1104 πŸ” 501 πŸ’¬ 9 πŸ“Œ 14

The safety of modern civilization, for better or worse, depends on anonymity due to how easily information is found nowadays. If society loses this, if it loses section 230, effective public outreach will be decimated. We cannot let these bills pass if we want to be able to keep our freedoms.

CALL

05.03.2026 22:33 πŸ‘ 70 πŸ” 83 πŸ’¬ 0 πŸ“Œ 2

Sadly this is a bi-partisan effort being led by the Democrats. We need to call ALL lawmakers and tell the Democrats (Warner, Blumenthal, etc) to stop voting for mass AI surveillance and censorship

05.03.2026 19:12 πŸ‘ 200 πŸ” 50 πŸ’¬ 4 πŸ“Œ 0

TTRPG SAFETY ANNOUNCEMENT:
CW: CSA materials
I've just heard from a trusted source (with evidence linked below)
that @ colbosch. bsky. social was arrested in June last year for possession of CSAM.
Sharing this info so that others can look into it for themselves and keep themselves safe.

05.03.2026 15:47 πŸ‘ 351 πŸ” 264 πŸ’¬ 25 πŸ“Œ 56
Preview
Censorship Does Not Keep Kids Safe KOSA is an alarming bill that would enable censorship and violate young people’s First Amendment right to freely access information. We must call on Congress to act now and protect free speech online.

Congress is taking up an alarming bill that would empower the government to decide what content is deemed dangerous to young people, encouraging censorship of content it doesn’t like.

Tell your members to support free speech and vote no on censorship.

05.03.2026 17:30 πŸ‘ 438 πŸ” 283 πŸ’¬ 7 πŸ“Œ 11

Text NO WAR to 30403 to call your Senator and tell them to also vote yes!!

04.03.2026 20:19 πŸ‘ 51 πŸ” 21 πŸ’¬ 0 πŸ“Œ 0

Did my calls yesterday and giving another today! Please please please call your reps <3 thank you

04.03.2026 23:56 πŸ‘ 37 πŸ” 30 πŸ’¬ 1 πŸ“Œ 0

How is it legal to sell these?

05.03.2026 00:37 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

DO NOT COMPLY.

DO NOT OUT PEOPLE, ESPECIALLY VULNERABLE CHILDREN

04.03.2026 17:29 πŸ‘ 915 πŸ” 521 πŸ’¬ 9 πŸ“Œ 3
YOU CAN FIGHT ICE
BY ANNOYING THEM
- MESS UP THEIR FOOD/DRINK ORDERS
- LOSE THEIR TICKETS/RESERVATIONS
- OVERTIGHTEN LUG NUTS, STRIP SPARK PLUGS
- GIVE THEM INCORRECT BUT REAL-SOUNDING
INFORMATION
- RECORD THEM WITH YOUR PHONES
- DISRUPT THEM WITH LOUD NOISES (CAR ALARMS, MUSIC, BANGING METAL) AND BRIGHT LIGHTS (FLASHLIGHTS, HEADLIGHTS)
- SHAME THEM
Fuck yeah
THE FASCISTS WANT COMPLACENCY.
DO NOT GIVE IT TO THEM.
Fuck them White supremacist
IT IS YOUR RIGHT TO MAKE Β«MISTAKES"

YOU CAN FIGHT ICE BY ANNOYING THEM - MESS UP THEIR FOOD/DRINK ORDERS - LOSE THEIR TICKETS/RESERVATIONS - OVERTIGHTEN LUG NUTS, STRIP SPARK PLUGS - GIVE THEM INCORRECT BUT REAL-SOUNDING INFORMATION - RECORD THEM WITH YOUR PHONES - DISRUPT THEM WITH LOUD NOISES (CAR ALARMS, MUSIC, BANGING METAL) AND BRIGHT LIGHTS (FLASHLIGHTS, HEADLIGHTS) - SHAME THEM Fuck yeah THE FASCISTS WANT COMPLACENCY. DO NOT GIVE IT TO THEM. Fuck them White supremacist IT IS YOUR RIGHT TO MAKE Β«MISTAKES"

Monkey wrenching is such an under appreciated form of protest

03.03.2026 21:20 πŸ‘ 4098 πŸ” 1697 πŸ’¬ 30 πŸ“Œ 19

One of my favorite parts of last night’s election results.

The only Democratic talking point should be: we don’t abandon anyone.

04.03.2026 20:32 πŸ‘ 1751 πŸ” 411 πŸ’¬ 16 πŸ“Œ 8

THE FIGHT IS NOT DONE (Contact resources in the reply to the post)

MAKE YOURSELF HEARD

03.03.2026 23:57 πŸ‘ 327 πŸ” 333 πŸ’¬ 2 πŸ“Œ 0

The end.

04.03.2026 00:43 πŸ‘ 2 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0

I looked at the trailer, and it feels like it came straight from 2016.

04.03.2026 00:32 πŸ‘ 1 πŸ” 0 πŸ’¬ 0 πŸ“Œ 0
Preview
We've freed Cookie's Bustle from copyright hell. Here's how. | Video Game History Foundation We've stopped a persistent copyright troll who was trying to impede our work. Here's everything we learnedβ€”and how we beat them.

BREAKING: We've freed Cookie.

Following an investigation by VGHF, Ukie and Web Capio have suspended DMCA takedowns for Cookie's Bustle on behalf of Graceware, SL.

More info:

03.03.2026 17:57 πŸ‘ 6346 πŸ” 3179 πŸ’¬ 81 πŸ“Œ 308