Conan (@_Conan0x) header
Conan (@_Conan0x) profile picture

poasting

🌴USA🌴
Joined May 2025
Adapt-1 TCP and vTranformer matched the ceiling, Mamba was close. But the sample budget is the story: Adapt-1 TCP: 510 online interactions per run vTransformer: 1,000,000 environment steps Mamba: 1,000,000 environment steps
A reward can tell an RL/online learner that something worked without telling it which combination of internal signals made it work. Today, we’re unlocking two learning rules in Adapt-1 Preview: Counterfactual Utility Plasticity (CUP) and Temporal Context Projection (TCP).
1
9
1
35
3,795
Click register and then use your email. The sign-in with Google isn’t working for some reason
3
91
Messing around with image gen to try an help understand the dense technical article REI just released. Very interesting stuff going on.
4
7
38
1,533
Instead of a one-size fits all LLM, imagine AI that's self-contained. It starts from 0 and is given data to reason over and learn from and runs the reasoning process through its own state and update rules. This means the AI becomes a domain specialist for your specific workflow.
🎙️ State of Agentic Coding with @mitsuhiko returns for Episode 7 In this pre-Fable 5 episode: - looping? - no more human-designed programming languages? - the state of local coding models - the marketing of token-maxing, and more
3
195
Yes, might be happening already, though I doubt it would be just an individual but another AI lab that isn't focused on the scaling paradigm so doesn't need the same exhaustive resources. teapawt.org/kimmonismus/status/206…
Google DeepMind published a 60-page paper mapping the road from AGI to superintelligence, written by Hutter, Legg, and Genewein. No hype, just a sober analysis The paper uses three levels. AGI = roughly average human performance across most cognitive tasks. ASI = a system that beats large, well-coordinated groups of human experts across virtually everything (their bar: tens of thousands of experts working ten years on one problem). Universal AI / AIXI = the theoretical ceiling, uncomputable, only approachable from below. Then they explore the question of how this could be achieved: Scaling compute, models, and data, the continuation of the trend that drove the breakthrough so far. It is the only path with historical data available for extrapolation. The core question: Does quantity transform into quality? Even if individual models plateau, the sheer act of running millions of faster AGI instances could trigger the leap. (A quick aside: that is a fascinating philosophical idea. It always reminds me of Hegel’s dialectic, the notion that quantity transforms into quality. We ought to start drawing on philosophical theories to make sense of the future.) Algorithmic paradigm shifts: a genuine break from the transformer pretraining paradigm. New architectures, new learning methods. However, hard to predict by definition. Recursive self-improvement: AI accelerates AI research, which produces better AI, which accelerates research further. Multi-agent coordination: superintelligence emerges from large collectives of AGI agents working together, like automated corporations or AI economies. Collective intelligence potentially far exceeding any individual model. The authors naturally point to what I repeatedly describe as the biggest bottleneck: energy. I recently linked to a few graphs showing, on the one hand, the extent to which energy is already becoming a problem and, on the other, how China dominates the expansion of both nuclear and solar energy in the global race. But the authors also address a profound shift in the world of work in a post-AGI era. I would say this is a reality we must face. So, it is not just about scaling, but also about whether the underlying conditions - such as energy and hardware - can be effectively established. Six things that could slow or stop all of this: The data wall. Quality training data runs out, possibly before the end of this decade. Resource demand grows too fast. Energy, chips, rare earths, investment. The physical infrastructure can't scale arbitrarily. The neural paradigm hits a ceiling. Pretrained transformers plus fine-tuning may not be enough to reach AGI, let alone go beyond it. Research gets harder. Keeping Moore's law going already needs 18x more researchers than in the 1970s. Ideas are genuinely harder to find as fields mature. The abstraction barrier. Models trained on human concepts may never invent new ones from scratch. Saturating GPQA or SWE-bench shows mastery of what humans already worked out, not the ability to go beyond it. Train only on pre-Newtonian physics and you won't reason your way to relativity. Deliberate slowdown. Regulation, accidents, public backlash. Real, but likely countered by the competitive pressure between companies and nations. I think it’s great that Google is addressing questions such as which paths they believe lead to AGI, what the road to ASI might look like, what challenges will arise, and much more. Overall, however, it sounds to me like all of this could actually succeed, making it, in that sense, a call to discuss and reflect on the consequences.
3
78
grei
Awesome new theme issue of Philosophical Transactions: ‘World models in natural and artificial intelligence’ Thank you @adamsafron! royalsocietypublishing.org/r…
1
5
165
Replying to @pmarca
Can they learn at inference time yet or evolve their architecture? How do they handle non-linguistic data like sensor data?
1
1
7
575