ทวีตที่ปักหมุด
My second short story release in English is ready: Tales of Illustrious Computer Scientists: Iola Varga, nun and computer scientist. invece.org/iola.html
English
antirez
41.7K posts
antirez
@antirez
Reproducible bugs are candies. I like programming too much for not liking automatic programming.
Sicily, Italy เข้าร่วม Mayıs 2007
762 กำลังติดตาม49.6K ผู้ติดตาม
I'm developing a new Redis data type: even after working for 1 month to the text specification, I had to make choices in every code detail possible. Otherwise the code could work? Yes. But would be truly high quality stuff? Nope. Today it is like that. Tomorrow we will see.
English
Wow, I totally disagree with this statement. At the current state, AI actually amplifies the developer to developer difference. If you were a 10x developer, you had good ideas + architectural clarity, this is a brutal advantage when using AI. Steering is a fundamental part of today's AI development.
Rohan Paul@rohanpaul_ai
Chamath on how AI agents are making the "10x engineer" distinction disappear because the most efficient "code paths" are now obvious to everyone. Just as AI solved chess and removed the mystery of the best move, AI is doing the same for coding, making the process reductive and removing technical differentiation. "I'm going to say something controversial: I don't think developers anymore have good judgment. Developers get to the answer, or they don't get to the answer, and that's what agents have done. The 10x engineer used to have better judgment than the 1x engineer, but by making everybody a 10x engineer, you're taking judgment away. You're taking code paths that are now obvious and making them available to everybody. It's effectively like what happened in chess: an AI created a solver so everybody understood the most efficient path in every single spot to do the most EV-positive (expected value positive) thing. Coding is very similar in that way; you can reduce it and view it very reductively, so there is no differentiation in code." --- From @theallinpod YT channel (link in comment)
English
According to rumors it was not licensed, which is odd given K2.5 modified MIT license (I talked about it in my YouTube channel) and the Cursor numbers. Yet Moonshot AI employees tweets confirming it is not licensed were deleted: not accurate or the company don't want to get into fights?
Fynn@fynnso
was messing with the OpenAI base URL in Cursor and caught this accounts/anysphere/models/kimi-k2p5-rl-0317-s515-fast so composer 2 is just Kimi K2.5 with RL at least rename the model ID
English
Mario can tell you: in Austria it is *even* worse, it's not an Italian problem. My brother @DarioSanfilippo moved to Austria to find a system that is at least as bad, from this point of view. The commission knows about those small mafias of quasi-mandatory financial advisors and so forth.
English
@matteocollina @badlogicgames @lucalanziani Maybe they will do a mess, but in the current proposal it is documented that you *jump* the single state procedures with this company form. I believe that the Commission point is exactly that, they know that for local interests they can't win the local battles, so they bypass it.
English
@antirez You plan to automatically cut off anyone who is starting right now.
Not an idea I can endorse.
English
Clearly we need some kind of approval process to avoid the ability to open PRs for new accounts without a past GitHub contribution history and so forth. Here the problem is not LLM assisted programming per se, but agents around on the internet doing shit like that.
clem 🤗@ClementDelangue
Our biggest open-source repos are getting overwhelmed by AI slop which literally makes Github unusable (~a new pull request every 3 minutes). Fun new challenges in an agentic world!
English
@BlackOpsREPL Just require approval for accounts created >= 2023 to start. It's not hard to come up with something. The problem is now that GitHub has the times of megacorps.
English
@antirez There is already a whole scene of fake accounts with generated histories from since LLMs came out, so ~2023.
English
antirez รีทวีตแล้ว
We are introducing EU Inc. To make building and growing a business across the EU faster, simpler, and smarter.
🔸 Start a company in less than 48 hours
🔸 No minimum capital requirement
🔸 Fully online and borderless
English
Cito dal video da cui è stata estrapolata questa affermazione: "allora togliamo anche il latino, il greco, la matematica di ordine superiore" (dalle scuole). Il punto non è il latino in sé, ma spingere gli studenti a fare cose complicate, qualsiasi esse siano, linguistiche, logiche, eccetera, a prescindere da quanto sembrino utili poi nella pratica.
Italiano
Italiano
Il costante calo delle iscrizioni agli studi classici nelle scuole superiori italiane ha alimentato un dibattito sulla necessità di una riforma sostanziale di questo curriculum. A difesa del liceo classico, molti hanno sostenuto che esso aumenti l’apertura mentale 1/3
Benedetta Frucci@BenedettaFrucci
Sostenere che lo studio del latino sia inutile nel tempo dell’intelligenza artificiale non solo è sbagliato: è esattamente l’opposto. 1) Lo studio del latino sviluppa la capacità di analisi sintattica, la capacità di distinguere funzione e forma, l’attenzione alla gerarchia delle informazioni. Sono le stesse competenze richieste per comprendere e costruire sistemi complessi, inclusi quelli algoritmici. 2) Richiede disciplina mentale e insegna a dubitare: nella traduzione si scompone la frase, la si ricostruisce, si verifica la coerenza e il significato. Tradurre è un esercizio di pensiero critico, fondamentale quando si usa una AI. 3) Molti termini scientifici, giuridici e tecnologici derivano dal latino. Capire l’etimologia significa cogliere sfumature concettuali, migliorare la qualità del linguaggio. Le materie umanistiche saranno sempre di più un bene preziosissimo. Ciò che sarà prodotto dalla mente umana sarà sempre più raro e per questo avrà un valore sempre più alto. Mettere in contrapposizione il latino e l’AI è profondamente miope. Questa è la dimostrazione che i grandi esperti di economia (o presunti tali) faticano a capire la politica. E il futuro. Perchè il latino ci educa all’intelligenza artificiale molto più di qualche professore in cerca d’autore
Italiano
antirez รีทวีตแล้ว
🔥 Meet Mistral Small 4: One model to do it all.
⚡ 128 experts, 119B total parameters, 256k context window
⚡ Configurable Reasoning
⚡ Apache 2.0
⚡ 40% faster, 3x more throughput
Our first model to unify the capabilities of our flagship models into a single, versatile model.
English
@rowantrollope @elianto_nL When there is to create some new system, a good starting point could be to provide both a description of the new system *and* old code: together they specify a lot of things that can completely steer the new system code generation.
English
This. When creating a new system, you write the initial specification, composed of both a high level human language description, and even description of the exact algorithms, data structures, when it matters (I go to the level of outline the C struct and functions for Redis work). Then I iterate with the LLM, updating the spec, creating the code. The full picture is captured by both the code and the implementation itself: they say different levels of things.
English
@antirez Smart developers already work this way. Intent in plain language, implementation in proper code. Best of both worlds.
English