Sam Altman:"Scaling LLMs won't get us to AGI" – maybe we found a path
Hi HN, At the 2023 Hawking Fellowship at Cambridge Union, a student asked Sam Altman: “To get to AGI, can we just keep min-maxing language models, or is there another breakthrough that we haven’t really found yet?” Altman’s response was telling: “We need another breakthrough… I don’t think that doing that will get us to AGI. If, for example, superintelligence can’t discover novel physics, I don’t think it’s a superintelligence. Teaching it to clone the behavior of humans and human text – I don’t think that’s going to get there.” If Altman is right that scaling alone is a dead end, then this…
What it does
In the maker’s words, at launch
Hi HN, At the 2023 Hawking Fellowship at Cambridge Union, a student asked Sam Altman: “To get to AGI, can we just keep min-maxing language models, or is there another breakthrough that we haven’t really found yet?” Altman’s response was telling: “We need another breakthrough… I don’t think that doing that will get us to AGI. If, for example, superintelligence can’t discover novel physics, I don’t think it’s a superintelligence. Teaching it to clone the behavior of humans and human text – I don’t think that’s going to get there.” If Altman is right that scaling alone is a dead end, then this might be the beginning of the real race - not just to build bigger LLMs, but to invent the architectures that can discover the unknown. This echoes a long-standing question in AI research: what lies beyond scaling language models in the pursuit of true general intelligence? Did we finally stumbled onto the next breakthrough Altman was hinting at? We think so. Collective AGI: The Civilizational Path to AGI Rooted in History: Human intelligence didn’t become powerful in isolation. A single brain is limited. What made human intelligence general and world-changing was the civilizational process - networking, collaboration, culture, institutions, governance, commerce, law, ethics - compounding across generations. A Pattern, a Fractal: Ecosystems evolve through interdependent species; human intelligence evolved through interdependent minds. Civilization was humanity’s first great outcome of collective intelligence, recursively amplifying the reach of individual cognition. Applied to AGI: In the same way, AGI won’t emerge from a single artifact or mono-worldview. It must arise through plurality of AI forms, multi-agent networks, evolving institutions, and shared participation - the same mechanisms that scaled human intelligence, now applied to artificial intelligence. If that’s true, the next breakthrough isn’t bigger LLMs. It’s building civilizational ecosystems for AI societies scale - that don’t just predict text, but evolve new knowledge, new institutions, and new ways of thinking. I'm Kanishka Nithin, I’ve been working on AI for over a decade, and today I’m excited (and a little nervous) to share something deeply personal and long in the making: AGI Grid - our open effort to together build what we call Collective AGI - The open, fastest, safest, and most efficient path to Artificial General Intelligence. AGI Grid: It is a open civilizational infrastructure ecosystem for collective AGI Website: https://www.agigr.id Vision paper: https://resources.agigr.id The collective AGI ecosystem consists of 12 Open source projects - listed at Docs: https://docs.agigr.id What we’d love from HN: Feedback: Does this framing of Collective AGI resonate? Critique: Where do you see the holes? Collaboration: If you’re working on agents, collective intelligence, or diverse cognitive architectures, let’s talk. This project grew out of both crisis and conviction, and we believe the next leap in AI won’t come from a single giant model, but from grids of diverse forms of intelligence, aligned and cooperating. -Team AGI Grid: A Collective AGI
Does a similar job
all alternatives →
Getting GLM 5.2 running on my slow computerJul 2026 · github.com · ▲937A few days ago I found myself trying out GLM 5.2 and was really positively impressed. The capabilities and security I was getting from this LLM are similar to those I've gotten from models like Claude or GPT, and this really surprised me. But then I thought, "I wonder how it would work on a normal computer like mine," and above all, "I wonder if it would work without going into OOM on a computer like mine." So I started working with the help of agents to test this possibility. I started converting the model to int4, understanding MTP usage, and if possible implementing DSA for long context.…
- L3Llama 3.2 Interpretability with Sparse Autoencoders2024 · github.com · ▲579
I spent a lot of time and money on this rather big side project of mine that attempts to replicate the mechanistic interpretability research on proprietary LLMs that was quite popular this year and produced great research papers by Anthropic [1], OpenAI [2] and Deepmind [3]. I am quite proud of this project and since I consider myself the target audience for HackerNews did I think that maybe some of you would appreciate this open research replication as well. Happy to answer any questions or face any feedback. Cheers [1]…
- IBI built a tiny LLM to demystify how language models workApr 2026 · github.com · ▲915
Built a ~9M param LLM from scratch to understand how they actually work. Vanilla transformer, 60K synthetic conversations, ~130 lines of PyTorch. Trains in 5 min on a free Colab T4. The fish thinks the meaning of life is food. Fork it and swap the personality for your own character.
- WMWe made glhf.chat – run almost any open-source LLM, including 405B2024 · glhf.chat · ▲161
Try it out! https://glhf.chat/ Hey HN! We’ve been working for the past few months on a website to let you easily run (almost) any open-source LLM on autoscaling GPU clusters. It’s free for now while we figure out how to price it, but we expect to be cheaper than most GPU offerings since we can run the models multi-tenant. Unlike Together AI, Fireworks, etc, we’ll run any model that the open-source vLLM project supports: we don’t have a hardcoded list. If you want a specific model or finetune, you don’t have to ask us for it: you can just paste the Hugging Face link in and…
- AHAGI Hits a Structural Wall – A Billion-Dollar Problem2025 · ▲8
This paper formally defines where current AGI hits a structural wall — not a technical one. It shows that no amount of scaling, reinforcement learning, or recursive optimization will break through three deep epistemological and formal constraints: 1. Semantic Closure — An AI system cannot generate outputs that require meaning beyond its internal frame. 2. Non-Computability of Frame Innovation — New cognitive structures cannot be computed from within an existing one. 3. Statistical Breakdown in Open Worlds — Probabilistic inference collapses in environments with heavy-tailed uncertainty.…
More life & fun this month
the category →- TL
Life & fun · 11d ago · louisabraham.github.io
Photosynthesis fires two of your iPhone
Life & fun · 29d ago · photosynthesis.camera
- CCCreatium Coach▲320
Your multimedia mentor that takes you from mid to great
Life & fun · 11d ago · producthunt.creatium.info
SoloUno▲310Take control of hair pulling, nail biting & skin picking
Life & fun · 29d ago · solouno.io
Launched alongside, August 2025
the whole month →
- IS
I built the world's most impractical 1000-pixel display and anyone in the world can draw on it. It draws a single pixel at a time and takes 30-60 minutes to complete a single image. Anyone can participate in the project by voting for the next image to be drawn, and submitting images. https://kilopx.com/
Work · 2025 · benholmen.com
- KT
Kitten TTS is an open-source series of tiny and expressive text-to-speech models for on-device applications. We are excited to launch a preview of our smallest model, which is less than 25 MB. This model has 15M parameters. This release supports English text-to-speech applications in eight voices: four male and four female. The model is quantized to int8 + fp16, and it uses onnx for runtime. The model is designed to run literally anywhere eg. raspberry pi, low-end smartphones, wearables, browsers etc. No GPU required! We're releasing this to give early users a sense of the latency and voices…
Dev tools · 2025 · github.com
- IW
I was wondering how I can arrange objects along a spherical helix path, and read some articles on it. I ended up learning about parametric equations again, and make this visualization to document what I learned: https://visualrambling.space/moving-objects-in-3d/ feel free to visit and let me know what you think!
Life & fun · 2025 · visualrambling.space
- TC
For HTML Day 2025 [1], I made a web service that displays the current sky at your approximate location as a CSS gradient. Colours are simulated on-demand using atmospheric absorption and scattering coefficients. Updates every minute, without the use of client-side JavaScript. Source code and additional information is available on GitHub: https://github.com/dnlzro/horizon [1] https://html.energy/html-day/2025/index.html
Dev tools · 2025 · sky.dlazaro.ca