trinidad matta

law student. science communicator.

[Tib3rius @0xTib3rius]

Oh.

[Justine Moore @venturetwins]

I just got bullied by AGI

[natochi @natochi_]

🇲🇽🇲🇽 MÉXICOOOO, ESTAS LISTO?

En pocas horas cerramos la postulaciónes para la Platanus Hack 26: CDMX, ft. @firecrawl.

Serán 100 de los mejores hackers de México construyendo producto por 36 horas, desde cero.

Alojando dos noches en las oficinas de @fintual. Comida, energética, y taquitos.

Postula junto a tu team en el link de los comentarios! 👇

[Claude @claudeai]

Releasing a model this capable comes with risks. Without safeguards, Fable 5’s capabilities in areas like cybersecurity could be misused to cause serious damage.

Queries on a narrow range of topics will instead receive a response from our next-most-capable model, Opus 4.8.

[trinidad matta @tonchimatta]

pasó una semana y me veo obligada a volver a postear esto

[Rob Hallam @robj3d3]

It's official, we have AGI.

[Radio ADN @adnradiochile]

Inédito fallo de la Corte Suprema: suspenden a abogada que inventó citas usando inteligencia artificial

https://t.co/X20AqGgVoZ

[trinidad matta @tonchimatta]

@La_Segunda Ser embajador de anthropic es ser un líder comunitario de difusión. No da ninguna autoridad ni es un empleado de anthropic como para estar diciendo eso. Si el no quiere emplear ingecos (a pesar de serlo) es tema suyo. No habla en representacion de nadie. https://t.co/1ZNghi3ggI

[trinidad matta @tonchimatta]

@RatOrthodox @fdosmither @feelsdesperate Important to consider that it played a huuuge role in the economic crisis of 1982. Unemployment rose to 23% and the GDP fell by 14% that year.

It sure had great things, but it wasn’t a miracle either

[Suhail Kakar @SuhailKakar]

@claudeai

[trinidad matta @tonchimatta]

spiders in the web

[kai @ssvankai]

accidentally stumbled upon this crazy paper abstract

[Annie? Rauwerda? @anniierau]

"why use google?" webpage from 1999

[natochi @natochi_]

Vamos a enseñar a 150 personas a construir su primera app con IA.

Me refiero a abogados, profesores, medicos. Personas que todavía no están usando ni siquiera chatgpt.

Va a ser en vivo, completamente en inglés, en la Universidad de Chile, patrocinado por la Cámara Chilena de la Construcción.

Si conoces a alguien a quien le podría entregar valor, invítala/o.

Además, va a haber una sesión para que cada quien aplique lo que vio con créditos de IA gratis (~50 dólares en créditos).@ @

Jueves 4 de junio, ~4pm
Universidad de Chile, FEN.

https://t.co/jBYqhLT3P9

Hey, it's going to be in English. Even if you haven't used it since high school, it's a good opportunity to practice your listening, no? You might just leave with your first app.

[sid @immasiddx]

Hear me out too

[AA @measure_plan]

i made snake but the only way to move is by playing guitar chords

[trinidad matta @tonchimatta]

As a double space fan, I hate to see AI come for my style of writing.

Yes, I’ll still do it, i don’t care.

[AnonTeacher @teacherhottakes]

@Freyy_is It's not x, but y.

Not this partial thought. Not this fragment. Not this half idea.

It's the "quietly" doing any action.

It's the double space between each line of a post.

Not this. Not that. Not the other.

It's all of it. And the worst part? It's EVERYWHERE.

[AnonTeacher @teacherhottakes]

@Freyy_is It's not x, but y.

Not this partial thought. Not this fragment. Not this half idea.

It's the "quietly" doing any action.

It's the double space between each line of a post.

Not this. Not that. Not the other.

It's all of it. And the worst part? It's EVERYWHERE.

[Nav Toor @heynavtoor]

Researchers at EPFL proved your AI is lying to you.

Not sometimes. Most of the time.

They built one of the hardest hallucination tests ever made with Max Planck Institute. 950 questions. Four domains where being wrong actually hurts. Legal. Medical. Research. Coding.

Then they ran every top model on it.

The results.

GPT-5. Wrong 71.8% of the time.

Claude Opus 4.5. Wrong 60% of the time.

Gemini 3 Pro. Wrong 61.9% of the time.

DeepSeek Reasoner. Wrong 76.8% of the time.

These are the smartest AI models on Earth. The ones you trust with your career. Your health. Your money.

You think turning on web search fixes it.

It doesn't.

Claude Opus 4.5 with web search. Still wrong 30.2% of the time.

GPT-5.2 thinking with web search. Still wrong 38.2% of the time.

The internet attached. Still lying to you in 1 out of every 3 answers.

Now the part that should scare you.

Medical questions. The one place being wrong can kill you.

GPT-5 hallucinated 92.8% of the time on medical guidelines.

Claude Haiku 4.5 hallucinated 95.7% of the time.

Gemini 3 Flash hallucinated 89% of the time.

Nine out of ten medical answers from popular AI models. Wrong.

It gets worse.

The longer you talk to it, the more it lies.

Early mistakes cascade. The model starts citing its own earlier hallucinations as facts. Your third message is more wrong than your first.

The paper, in its own words: "hallucinations remain substantial even with web search."

This is what hundreds of millions of people are doing right now. Asking software that lies in the majority of its answers. About their health. About their job. About their legal case. About their code.

Most are not checking.

Most never will.

But please. Keep using ChatGPT for medical advice.

The doctors need a break.

https://t.co/dHBP5CDpTM

[Peyman Milanfar @docmilanfar]

being wrong is the best way to get cited a lot

[Mushtaq Bilal, PhD @MushtaqBilalPhD]

Sci-Hub is an evil website that pirated 85M+ research papers and made them freely available

And now they've added AI to their database to make Sci-Bot.

It answers your questions using latest, full-text articles.

But DO NOT use it. We should all try to make billion-dollar academic publishers richer.

I'm putting the link below so you know how to avoid it.