Daniel Mauricio Flores 🐲 @devniel

There exists a martyrdom of those who offer their lives in a single moment, and the more hidden martyrdom of those who give their lives day after day. This is exactly what all Christians are called to do: to give our lives with love, every day.

Crítica de Benedicto XVI al capitalismo salvaje y su idolatría del lucro:

The rollout of Pope Leo’s AI Encyclical, Magnifica Humanitas (Magnificent Humanity), was the most underreported story of the year. He personally presented the document at a Vatican press conference, an unusual move for a Pope. He knew the significance of AI.

anna artreides @annapdal

a week ago

A única pessoa que vi fazendo discurso decente sobre IA realmente é o papa. Ele não diz só "não use" "Use de maneira de que se a IA desaparecer amanhã, vc ainda saiba pensar... criar, existir."

42 7K 42K 360K 3K

“We live in the age of search engines that offer us thousands of instantaneous answers. Yet, no matter how powerful they may be, they merely return what already exists and tend to give us what we already want. In the age of technology and consumerism, the great danger is not only censorship but also homogenization, and not merely imposed silence but the noise that makes it impossible to listen. “Art is a different kind of ‘search engine.’ It does not proceed by indexing but by inquiry. Indeed, its inquiry is a focused attention that is not satisfied with the mere sum of facts, for it never ceases to dig deeper, seeking the thread that unites them. Your art, when it resists homogenization, when it draws our gaze, when it compels us to pause before that which cannot be consumed, is a form of cultural and spiritual regeneration.” — Pope Leo XIV

Explaining the Catholic Church to Godless AI researchers: Imagine a lab thats spent 2000 years working on alignment

The year is 2026. San Francisco tells the Vatican it has the power of God to put a soul in a computer program. The last Crusade begins.

A única pessoa que vi fazendo discurso decente sobre IA realmente é o papa. Ele não diz só "não use" "Use de maneira de que se a IA desaparecer amanhã, vc ainda saiba pensar... criar, existir."

Christopher Hale @ChristopherHale

a week ago

“AI can process information quickly, but it cannot replace human intelligence. And don’t ask it to do your homework for you.” — Pope Leo XIV

81 3K 17K 1.2M 2K

Pope Leo XIV at the encyclical launch on May 25: “Artificial intelligence needs to be disarmed.”

I’m enjoying two things: 1. The Pope, a deeply educated man in mathematics and the humanities, crafts a thoughtful warning about the delicate balance between human creativity and machines. 2. It’s posted from the account specifically dedicated to messaging in Latin, and folks are losing their minds. He’s on another level, and the tech world is not ready for it.

Leo Papa XIV @Pontifex_ln

a week ago

Hac intellegentiae artificialis aetate, urgens fit humanam artem ab iis distinguere, quae machinis efficiuntur. Ars enim et ea, quae machina ex innumeris alienis imaginibus statisticae ope computationis generare potest, ontologice, prius etiam quam aesthetice, inter se differunt.

282 6K 36K 2.7M 4K

“AI can process information quickly, but it cannot replace human intelligence. And don’t ask it to do your homework for you.” — Pope Leo XIV

Prediction: organised religion is going to come roaring back as a response to loss of meaning in post AI world. The catholics recognise this and are getting ahead of the game

Christopher Hale @ChristopherHale

a week ago

“AI can process information quickly, but it cannot replace human intelligence. And don’t ask it to do your homework for you.” — Pope Leo XIV

81 3K 17K 1.2M 2K

We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for aerospace maintenance documentation. LLMs well-versed in this language and it comes with heavy constraints on clean writing style that I often find a lot more readable. Sometimes I've tried to soften it a bit e.g. ask for "80% of the way to ASD-STE100" because the spec is quite stringent. But even better: Diagrams / images. Instead of writing, ask your LLM to create a diagram. These can be a lot easier to process, parse, and understand. But even better: Web pages. Ask for output "in HTML" to get a beautiful, interactive webpage. LLMs are getting really good at frontend and can create beautiful experiences, animations, etc. But even better: Explainer videos. The output format I am most bullish on is fully custom / bespoke explainer videos generated on any arbitrary topic. Experiment with things like "Create a 3b1b style video explainer on X. Use my ElevenLabs API key for audio narration". (you'd need an API key for the latter or you can ask your LLM to find you decent free alternatives that use your local compute). This is actually starting to work! In summary: - As LLMs get better, they will do more and more of the legwork autonomously, and a lot more of our work will rise up the abstractions into oversight and understanding. - Luckily, LLMs can help here too because as intelligence and code are increasingly abundant, you can ask for large, custom, discardable software artifacts (e.g. web apps, video explainers) that would have never made sense to create before. Push the boundaries here and you'll be surprised.

Aviation solved AI slop in 1986. Karpathy is telling people to ask their LLM to write in ASD-STE100, a controlled language built for aircraft maintenance manuals. I went down the rabbit hole on where this spec comes from, and the backstory makes the hack even better. Back in the late 1970s, European airlines were handing English manuals to mechanics around the world who mostly didn't speak English as a first language. One ambiguous sentence in a fuel line procedure can get people killed. So the industry built a version of English where ambiguity is banned at the dictionary level. The spec allows roughly 900 approved words. Each word gets one meaning and one part of speech. "Close" can only be a verb, because "close the door" and "the close door" can't be allowed to coexist in a repair manual. "Follow" only means to come after. If you want obey, you write obey. Procedural sentences max out at 20 words. One instruction per sentence. Active voice only. Around 1,200 common words are explicitly banned, each with a mandated replacement. You don't "commence pumping." You start the pump. English has about 170,000 words in current use. Aviation decided mechanics get 900, and planes became safer to maintain because of it. Now flip it to LLMs. Models hedge and synonym-cycle because the training data rewards sounding fluent. STE is 40 years of accumulated rules for stripping exactly that out, written by people whose readers would die if a sentence could be read two ways. It might be the strongest anti-slop prompt in existence, and it's a free PDF.

Andrej Karpathy @karpathy

2 weeks ago

We'll be spending a lot more time trying to understand the outputs of language models. A few thoughts, tips & tricks: Writing. Something I've had success with: Ask your LLM to explain something in ASD-STE100, it's a controlled language specification originally developed for

2K 7K 55K 7.8M 81K

We gave Jev 2,029 real phone calls. No transcripts or audio; it never heard a word. Our AI receptionist's calls were reduced to pure structure, meaning turns, tool calls, workflow stages and timing. During the calls, Jev made 38,012 turn-level forecasts at 118 ms median latency, reviewed every call with five typed questions and produced 10,145 answers in 26 seconds with 256 requests in flight. The experiment was zero-shot, with no fine-tuning or examples from our data. We compared Jev's forecasts with what actually happened in the EHR. By the halfway point, Jev could meaningfully separate calls that would book from those that wouldn't (AUC 0.78), and near the end it ranked them correctly 94% of the time. Even though Jev over-focused on visible errors our agent usually overcomes, it's still pretty incredible that it analyzed thousands of real calls in seconds for only $3.

it's f*cking gold TypeSafe just dropped the official skill that teaches Claude Code to build on Jev left alone, coding agents ask Jev one question per call and guess the field names. this skill rewires the order before a single line gets written: docs -> behaviour -> judgments -> one request -> code decides every rule in it, broken down on two pages: > see the failure behind each of its 11 instructions > pick Choice, Noul or Score in one glance > paste the prompt skeleton straight into your agent 12.2x the bill drop when 13 questions share one call, measured in TypeSafe's cookbook. a short file that does a lot of the thinking for your agent.

keyboard_arrow_left Previous keyboard_arrow_right Next

Mnimiy @Mnilax

3 weeks ago

it's f*cking insane Jev sat between GPT and me, killing every draft that broke my rules before i saw it. good setup. then Jev did not answer. the agent decided silence was safer and stopped sending me anything at all. took a second agent to unstick it. your checker needs a

7 15 204 265K 302

FFFFF: Focus. Move fast. Fail faster. Have fun. Never finish.

Introducing Unreal Agent: An open-source harness with state-of-the-art cost efficiency 39% cheaper than Codex+Astra on Terminal-Bench 4.0 while maintaining performance

i created a small directory for jev usecases.

first came jev. then laya. then laya-mlx. and now kev. all system one decision models. all within a span of 3 days. never seen a race this fierce before. let us breathe, folks. let us breathe.

@ctnicholasdev very cool !