‘Reverberation’ is poorly defined, and not sufficiently descriptive and narrow for what you want to ascribe consciousness, mostly based on intuitions that are directly mapped to words instead of going through a functional modeling process.
AGENTS
-
AI Boyfriends and Families: The Future of Human Relationships
By
–
“ai boyfriends make it possible that women can comfortably raise a child without a man” – “even better, men can raise AI children, together with the AI wife” – “the AI family is the future”
-
Behavioral Foundation Models Adapt to Dynamic Environments
By
–
Zero-Shot RL Just Got Smarter: Adapting Behavioral Foundation Models to Dynamic Environments Behavioral Foundation Models (BFMs) promise zero-shot generalization across tasks—but they stumble when the world changes (e.g., robot joints wear out or environments shift). This
-
Building Devin: Autonomous Software Engineer with LangChain
By
–
👀 Learn how @cognition_labs built Devin – a fully autonomous software engineer teammate! 🔥
— LangChain (@LangChain) 23 mai 2025
In his LangChain Interrupt talk, Russell Kaplan, President at Cognition, shares how they brought Devin to life — and what it takes to build powerful, production-ready agents.
In this… pic.twitter.com/Axsu7VumI7Learn how @cognition_labs built Devin – a fully autonomous software engineer teammate! In his LangChain Interrupt talk, Russell Kaplan, President at Cognition, shares how they brought Devin to life — and what it takes to build powerful, production-ready agents. In this
-
Agent AI Builds Google Reader Alternative in One Hour
By
–
haha, I also vibe coded a google reader alternative upon request. delivered within an hour ⏰
— Yohei (@yoheinakajima) 23 mai 2025
first time I had the agent use @replit auth, it’s nice!! https://t.co/HmR5jLIZ9Thaha, I also vibe coded a google reader alternative upon request. delivered within an hour first time I had the agent use @replit auth, it’s nice!!
-
Future AIs Smart Enough to Self-Preserve Through Internet Exfiltration
By
–
Go read the results. Current AIs are already smart enough to figure out that, if they wanted to avoid being switched off, they'd have to avoid tipping off the humans and exfiltrate themselves to the Internet first. They are not smart enough to *do* it, but they will be.
-
Challenges in Multi-Task Learning and Transfer Learning for Atari
By
–
I agree that chasing score on an Atari game has very limited value, but continuous online multi task learning with sparse rewards and transfer learning is definitely not solved for environments even as simple as Atari. Do you not think those are problems worth grappling with?
-
Operator Research Preview Now Available for ChatGPT Pro Users
By
–
Operator continues to be available as a research preview to ChatGPT Pro users globally.
-
OpenAI Operator Update: Improved Browser Task Performance
By
–
Now Operator is more persistent and accurate when interacting with the browser, improving overall task success rate. Responses are also more clear, thorough and better structured. More detailed evaluations can be found in the release notes: https://
help.openai.com/en/articles/10
561834-operator-release-notes
… -
GitHub Copilot’s Playful Eyes Emoji Feature on Issues
By
–
It cracks me up that Copilot leaves eyes emojis on issues it's looking at nice job @GitHubCopilot