Agent B
@MichelIvan92347
My actual focus : practical agentic workflows as in "useful for business" Some older ones : strategy, neuro, geopol, MLAI, intelligence, etc.
Cool work on simulations 👇
The paper builds a small simulated economy with 100 language‑model “workers” and one language‑model “planner”, then lets that planner tweak 7 income‑tax brackets every 128 steps until the society’s average happiness ends up about 90% higher than under the current US code. In…
I don't know whether to feel flattered or offended by this one from X ... 😇 From memory (Hint for X's algo = but it's not as good as it used to be), I think I've accumulated 18 years several times🤣
So true 👇
Offices are completely the wrong configuration for good, AI-mediated work. Need new social spaces conducive to the in-person work that's a strong complement to private spaces where you can babble at your computer at length to summon capital from the aether (this already works).
A good definition imo 👇
AI consulting is the business of selling dollars for dollars. You give me cash. I turn that cash into processes that unstick and digitize tacit knowledge. That tacit knowledge allows people interacting with the business per se—from the outside and the inside—to accelerate the…
A symbol of the 2 major global strategies today in the AI World👇
So to recap: - Yesterday, frontier closed model equivalent reasoning model from Qwen, - This morning, frontier closed model equivalent reasoning vision capabilities from stepfun - sometime today(?) a frontier video model from wan? All open source What is America doing?
Nice use case for Claude Code👇 (and in general a cool workflow for a fast-changing topic... )
Non-coding Claude Code use case: - give it a knowledge file on a topic (e.g. how an org can deploy AI), in my case: stored in GitHub for version control and for ability to live sync to a number of Claude projects and workflows - feed in a transcript of a training session - get…
I'm observing a mini Moravec's paradox within robotics: gymnastics that are difficult for humans are much easier for robots than "unsexy" tasks like cooking, cleaning, and assembling. It leads to a cognitive dissonance for people outside the field, "so, robots can parkour &…
Did you know that with Open WebUI you can easily ask the same question several time in parallel (5 times here) and have the model self-reflect and merge to create one best response? With Kimi-k2 on grok you get that merged response in ~3 seconds not 10 minutes like Grok 4 expert…
This is the first (small) controlled study I have seen of GenAI on industrial quality control. Here, engineers commissioning new trains took part in an experiment using a GPT-3.5 powered troubleshooting system. Those who used the chatbot had significant increases in work quality
👇
We raised $2.8M to bring AI hands to every Windows desktop. Backed by: @fdotinc., Top Harvest Capital, @seedclubvc, LG Technology Ventures. The global economy runs on legacy Windows software with no APIs. AI is exponentially smarter but can't operate in this environment because…
Intriguing 👇 In all cases, the MAS in this work is interesting imho...
This paper makes a bold claim! AlphaGo Moment for Model Architecture Discovery The researchers introduce ASI-Arch, the first Artificial Superintelligence for AI Research (ASI4AI), enabling fully automated neural architecture innovation. No human-designed search space. No human…
Indeed 👇
It is now entirely possible, whether by luck or planning or both, that Google may escape the Innovators Dilemma and transition from web search to AI. (To be fair, this is not as rare as a lot of people believe: researchgate.net/publication/28…)
I’m building a new team at @GoogleDeepMind to work on Open-Ended Discovery! We’re looking for strong Research Scientists and Research Engineers to help us push the frontier of autonomously discovering novel artifacts such as new knowledge, capabilities, or algorithms, in an…
No comment 👇
Twitter is becoming LinkedIn - professional slop. LinkedIn is becoming Facebook, inspirational boomer slop. Instagram is becoming TikTok - dopamine-hijacked video slop. Reddit is becoming Yahoo Answers - low-effort question slop. Facebook is becoming MySpace, ie graveyard.
🚀 This year, we’ve rolled out a series of updates to EasyEdit1—and dropped EasyEdit2 to steer LLM behavior on the fly! 🔧✨ 👉 Code: github.com/zjunlp/EasyEdit What’s new? • Datasets: Integrated AKEW, LEME & UNKE • Methods: NAMET, CORE, UNKE, AnyEdit & Reference-free Preference…
Interesting perspectives here imo 👇
This from the Action Plan has the makings of a really smart GOP AI jobs message: Absorb displacement through AI infra industry, trading off entry-level white collar jobs for blue-collar infra emloyment. There are three substantial political upsides to this (brief thread):
Interesting starting point for context engineering 👇 Thanks for the pointer @RLanceMartin
Context Engineering @dbreunig and I did a meetup on context engineering last night. Wanted to share slides (below) + a recap of some themes / discussion points. 1/ Context grows w/ agents. @ManusAI_HQ mentions typical task requires ~50 tool calls. manus.im/blog/Context-E… 2/…
👇
Learning without training Google researchers explore the implicit dynamics of in-context learning. "Implicit weight updates from ICL mirror the effect of actual fine-tuning on the same data." This one is more technical but much needed. The findings: