PUNTOGAP.COM

[OpenAI @OpenAI]

We’re updating GPT-5 Instant to better recognize and support people in moments of distress.

Sensitive parts of conversations will now route to GPT-5 Instant to quickly provide even more helpful responses. ChatGPT will continue to tell users what model is active when asked.

Starting to roll out to ChatGPT users today.

[Sam Altman @sama]

Sora update #1:

https://t.co/DC9ZpR7cSC

[Rohan Paul @rohanpaul_ai]

BRILLIANT @GoogleDeepMind research.

Even the best embeddings cannot represent all possible query-document combinations, which means some answers are mathematically impossible to recover.

Reveals a sharp truth, embedding models can only capture so many pairings, and beyond that, recall collapses no matter the data or tuning.

🧠 Key takeaway

Embeddings have a hard ceiling, set by dimension, on how many top‑k document combinations they can represent exactly.

They prove this with sign‑rank bounds, then show it empirically and with a simple natural‑language dataset where even strong models stay under 20% recall@100.

When queries force many combinations, single‑vector retrievers hit that ceiling, so other architectures are needed.

4096‑dim embeddings already break near 250M docs for top‑2 combinations, even in the best case.

🛠️ Practical Implications

For applications like search, recommendation, or retrieval-augmented generation, this means scaling up models or datasets alone will not fix recall gaps.

At large index sizes, even very high-dimensional embeddings fail to capture all combinations of relevant results.

So embeddings cannot work as the sole retrieval backbone. We will need hybrid setups, combining dense vectors with sparse methods, multi-vector models, or rerankers to patch the blind spots.

This shifts how we should design retrieval pipelines, treating embeddings as one useful tool but not a universal solution.

🧵 Read on 👇

[Rohan Paul @rohanpaul_ai]

This Github has a very wide collection of High-quality datasets, tools, and concepts for LLM fine-tuning.

All the datasets listed here should be under permissive licensing (Apache 2.0, MIT, cc-by-4.0, etc.).

Categorized into segments like Math & Logic, Code, Conversation & Role-Play, Agent & Function calling etc.

github. com/mlabonne/llm-datasets

[Fei Liu @feiliu_nlp]

Overwhelmed by the flood of papers? I highly recommend these must-read technical reports: K2, Qwen3, Qwen2.5-Omni, V3, and Claude 4. Huge thanks to the engineering teams who've done the heavy lifting, so we can learn from their insights!

- Kimi K2: https://t.co/JvxwAtjm97
- Qwen3: https://t.co/L3AnxfthTO
- Qwen2.5-Omni: https://t.co/aiIcpX6qZH
- Claude 4: https://t.co/EHwFrX0FcY
- DeepSeek V3: https://t.co/mi3wJUQqRh (Is the V3.1 report available yet?)

[DAIR.AI @dair_ai]

Top AI Papers of The Week (August 25-31):

- Memory-R1
- Anemoi Agent
- Jet-Nemotron
- Agentic Science
- Deep Think with Confidence
- Parallel Graph-Retrieval-Augmented Reasoning
- Fine-tuning LLM Agents without Fine-tuning LLMs

Read on for more:

[Rubén Darío Caramutti @RubenCaramutti]

Agree

[Qwen @Alibaba_Qwen]

How teams are using Qwen-Image in the wild →
🛒 With Qwen-Image and Qwen-VL, Alimama Creative turns plain product shots into high-converting posters.
Agents handle rewrites, prompts & visuals → fully automated creative pipeline.
From SKU → ad in seconds. ⚡

[Dan McAteer @daniel_mac8]

GPT-5 coding cheat sheet

from @OpenAIDevs

[Sam Altman @sama]

If you have been following the GPT-5 rollout, one thing you might be noticing is how much of an attachment some people have to specific AI models. It feels different and stronger than the kinds of attachment people have had to previous kinds of technology (and so suddenly deprecating old models that users depended on in their workflows was a mistake).

This is something we’ve been closely tracking for the past year or so but still hasn’t gotten much mainstream attention (other than when we released an update to GPT-4o that was too sycophantic).

(This is just my current thinking, and not yet an official OpenAI position.)

People have used technology including AI in self-destructive ways; if a user is in a mentally fragile state and prone to delusion, we do not want the AI to reinforce that. Most users can keep a clear line between reality and fiction or role-play, but a small percentage cannot. We value user freedom as a core principle, but we also feel responsible in how we introduce new technology with new risks.

Encouraging delusion in a user that is having trouble telling the difference between reality and fiction is an extreme case and it’s pretty clear what to do, but the concerns that worry me most are more subtle. There are going to be a lot of edge cases, and generally we plan to follow the principle of “treat adult users like adults”, which in some cases will include pushing back on users to ensure they are getting what they really want.

A lot of people effectively use ChatGPT as a sort of therapist or life coach, even if they wouldn’t describe it that way. This can be really good! A lot of people are getting value from it already today.

If people are getting good advice, leveling up toward their own goals, and their life satisfaction is increasing over years, we will be proud of making something genuinely helpful, even if they use and rely on ChatGPT a lot. If, on the other hand, users have a relationship with ChatGPT where they think they feel better after talking but they’re unknowingly nudged away from their longer term well-being (however they define it), that��s bad. It’s also bad, for example, if a user wants to use ChatGPT less and feels like they cannot.

I can imagine a future where a lot of people really trust ChatGPT’s advice for their most important decisions. Although that could be great, it makes me uneasy. But I expect that it is coming to some degree, and soon billions of people may be talking to an AI in this way. So we (we as in society, but also we as in OpenAI) have to figure out how to make it a big net positive.

There are several reasons I think we have a good shot at getting this right. We have much better tech to help us measure how we are doing than previous generations of technology had. For example, our product can talk to users to get a sense for how they are doing with their short- and long-term goals, we can explain sophisticated and nuanced issues to our models, and much more.

[Rubén Darío Caramutti @RubenCaramutti]

🔍 “Debemos saber. Sabremos.” — Hilbert, Amodei y la urgencia de comprender la IA

📍 En 1930, David Hilbert clausuró el Congreso Internacional de Matemáticos con una frase que resonó como un manifiesto de optimismo epistemológico:
“Wir müssen wissen, wir werden wissen.”
(“Debemos saber. Sabremos.”)

Esta declaración surgió en un contexto de profunda incertidumbre, marcado por las paradojas de Russell y los teoremas de incompletitud de Gödel, que sacudieron los cimientos de las matemáticas. 📐📚

🧠 Casi un siglo después, enfrentamos una crisis análoga: los modelos de inteligencia artificial —especialmente los basados en aprendizaje profundo— alcanzan niveles asombrosos de rendimiento, pero su funcionamiento interno permanece en gran parte como una “caja negra”.
⚠️ Esto plantea desafíos enormes en términos de confianza, seguridad y ética.

🔬 Dario Amodei, CEO de Anthropic, ha sido claro: necesitamos comprender y regular la IA antes de que alcance niveles de autonomía potencialmente incontrolables.
En su ensayo The Urgency of Interpretability (https://t.co/rp5k1WS4mX), advierte que mientras el poder de la IA se acelera, nuestra capacidad de interpretarla se está quedando atrás. 🧩

🔎 Por su parte, Demis Hassabis, CEO de Google DeepMind y galardonado con el Premio Nobel de Química 🏅 por AlphaFold, ha expresado preocupaciones similares (https://t.co/9gsIGI2VD1):
urge garantizar que comprendamos estos sistemas antes de que su autonomía supere nuestra capacidad de supervisión.

🔐 La interpretabilidad de los modelos de IA no es un simple desafío técnico. Es una cuestión fundamental para asegurar su alineación con valores humanos, y un requisito para su integración segura y ética en la sociedad.

🧭 Tal como en la época de Hilbert, cuando buscábamos fundamentos sólidos para las matemáticas, hoy debemos desentrañar el funcionamiento de los modelos de IA.
No sólo para mejorar su rendimiento, sino para asegurar que su evolución esté al servicio del bien común. 🌍🤝

✨ La frase de Hilbert vuelve a resonar como un faro para esta nueva era:

https://t.co/6zDE9InCgO

➡️ Debemos saber. Sabremos.

#IAResponsable #Interpretabilidad #TransparenciaAlgorítmica #Hilbert #DarioAmodei #DemisHassabis #DeepLearning #CajaNegra #FuturoDeLaIA #EpistemologíaDigital #LinkedInAcadémico #TechAndSociety #AIInterpretability #InteligenciaArtificial
#DebemosSaberSabremos

[Yann LeCun @ylecun]

Excellentes recommandations de la part de l'ancien ministre des finances Bruno Le Maire.
"Les Européens se trouvent devant une alternative simple : investir ou mourir. Investir dans l’IA pour améliorer notre productivité et donc offrir enfin de meilleurs salaires à ceux qui travaillent, investir dans la défense pour garantir notre sécurité,..."

[Arena.ai @arena]

New Arena launch: Sentiment Control - decoupling the impact of tone and emotion from response quality in human evaluation💗

How much do emojis, enthusiasm, and positive sentiment affect human preference? How can we adjust the leaderboard to counteract the effect of chat-optimized models with a friendly tone?🤔

Sentiment control answers this question by modeling the effect of sentiment on preference. After modeling this effect, we can adjust for it.💡

Early findings (Style + Sentiment control):
- Positive tone correlates with user preference!
- Claude-3.7-Sonnet, o1 improve in ranking under sentiment control
- Grok-3, Gemma-3, Llama-4-exp drop

More detailed analysis below👇

[Sam Altman @sama]

we have doubled rate limits for o3 and o4-mini-high for chatgpt plus subscribers.

enjoy!

[Rubén Darío Caramutti @RubenCaramutti]

This important event for our region in #VacaMuerta #Argentina is taking place today.
The #AI topic is more important than many of the topics on the current agenda. The #ClusterInfotech has organized an AI Hub to address the use of #ArtificialIntelligence in the business and entrepreneurial sectors of the region where #VacaMuerta is located.
If you want to watch it on YouTube, go to this link: https://t.co/AiOuXOMuPI…

[Rubén Darío Caramutti @RubenCaramutti]

Hoy se realiza este evento importante para nuestra región en #VacaMuerta #Argentina .
El tema #IA es más importante que muchos de los temas de la agenda de la actualidad. En el #ClusterInfotech se ha organizado un Hub IA donde se aborda la inserción de la #InteligenciaArtificial desde el sector de las empresas y emprendedores de la región donde se emplaza #VacaMuerta
Si lo quieren ver en Youtube entren a este enlace https://t.co/DSZyeq0sW1

[Rubén Darío Caramutti @RubenCaramutti]

#SMdayNQN

[PUNTOGAP.COM @PUNTOGAP_COM]

SMdayNQN

[neuq @NEUQUENCOMAR]

#SMdayNQN

[PUNTOGAP.COM @PUNTOGAP_COM]

The Big Bang no longer means what it used to https://t.co/cbCfEPcxYl

[PUNTOGAP.COM @PUNTOGAP_COM]

https://t.co/m9dfU7WRRd