This doesn't get enough attention: Tl;dr: we see the first self-improvements of the models (Meta, Anthropic), 100m context is not a technical problem and hallucinations can be partially bypassed.
RESEARCH
-

Neuromorphic Computing Framework: Scaling Time, Space, Energy
By
–
Neuromorphic Computing: A Theoretical Framework for Time, Space, and Energy Scaling Aimone et al.: https://
arxiv.org/abs/2507.17886 #ArtificialIntelligence #DeepLearning #MachineLearning -
Zuckerberg Announces Meta AI Push Toward Superintelligence and Self-Improvement
By
–
Quoi ?!
— VISION IA (@vision_ia) 31 juillet 2025
1. La superintelligence est en vue.
2. L’IA commence à s’améliorer elle-même.
3.Elle pourrait automatiser tout travail ayant de la valeur.
Après avoir volé tous les meilleurs employés de tous ces concurrents, Mark Zuckerberg annonce que Meta lance les… pic.twitter.com/GAInUeOxXuQuoi ?! 1. La superintelligence est en vue. 2. L’IA commence à s’améliorer elle-même. 3.Elle pourrait automatiser tout travail ayant de la valeur. Après avoir volé tous les meilleurs employés de tous ces concurrents, Mark Zuckerberg annonce que Meta lance les
-

Microsoft Study Reveals 40 Jobs Most Threatened by AI
By
–
Microsoft vient de publier une étude révélant les 40 métiers les plus menacés par l’IA… et les 40 métiers les plus sûrs face à cette révolution. Top 5 des métiers les plus à risque (Score IA le plus élevé)
1. Interprètes et traducteurs
2. Historiens
3. Agents de bord -

Neo AI launches NEO agentic machine learning engineer with sub-agents
By
–
Neo AI debuted NEO, an “agentic machine learning engineer” powered by 11 sub-agents
— The Rundown AI (@TheRundownAI) 31 juillet 2025
It runs autonomously, working like a full-stack engineer that handles everything from data exploration and feature engineering to training, deployment, and monitoringpic.twitter.com/ofFAtGLGMtNeo AI debuted NEO, an “agentic machine learning engineer” powered by 11 sub-agents It runs autonomously, working like a full-stack engineer that handles everything from data exploration and feature engineering to training, deployment, and monitoring
-

Google AlphaEarth Foundations Virtual Satellite AI System
By
–
Google introduced AlphaEarth Foundations, an AI that acts like a "virtual satellite"
— The Rundown AI (@TheRundownAI) 31 juillet 2025
It integrates satellite data into a single representation of Earth, giving a near-real-time view and helping with issues like crop health, deforestation, and morepic.twitter.com/icbkZu2tneGoogle introduced AlphaEarth Foundations, an AI that acts like a "virtual satellite" It integrates satellite data into a single representation of Earth, giving a near-real-time view and helping with issues like crop health, deforestation, and more
-
Zuckerberg Pursues Personal Superintelligence with Major AI Advances
By
–
AI NEWS: Mark Zuckerberg is gearing up to make "personal superintelligence" a reality. Plus, more news from OpenAI, Google, Neo AI, Showrunner, and Anthropic. Here's everything you need to know:
-
Horizon Alpha Model Performance: Quick Review and Comparison
By
–
Horizon Alpha – not bad honestly, for such a quick model – much better than Nectarine / Starfish; worse than Lobster. But better than most existing models at this task, e.g. better than Kimi K2, or GPT-4.1 https://t.co/NrLFTgkze6 pic.twitter.com/blK1GPoITJ
— Peter Gostev (@petergostev) 31 juillet 2025Horizon Alpha – not bad honestly, for such a quick model – much better than Nectarine / Starfish; worse than Lobster. But better than most existing models at this task, e.g. better than Kimi K2, or GPT-4.1
-

AI Leaders Report Signs of Self-Improvement in Systems
By
–
Lots of vague statements from leaders of the various AI labs about starting to see signs of self-improvement in AI systems (including Zuckerberg today), seems like proof that this is indeed happening would be pretty significant. (thanks o3 for providing details & saving me time)
-
Stanford NLP Researchers Honored with 25-Year Test-of-Time Award
By
–
I'm also proud to be a @stanfordnlp alumni 🙂 congrats to Gildea & @jurafsky on the well-deserved 25-year test-of-time award too!