My favorite depiction of utopia — LessWrong
For those who are trying to bring about a glorious transhuman utopia with the help of hopefully-aligned ASI, I think it's worth thinking explicitly a…
America Forever Bytes
LessWrong · AFBytes
Other
LessWrong
Recent coverage
The Origin of Uncertainty — LessWrong
When people ask what Fundamental Uncertainty is about, I usually say it’s a book about epistemology. If they want to know more, I say it’s a book arg…
Untitled Draft — LessWrong
TLDR: * Frontier models can detect when they're being evaluated and change their behavior, which risks compromising safety benchmarks. * We introdu…
Claude Opus 4.8: Capabilities and Reactions — LessWrong
You need a lot of data points to understand a new model, and what you have. …
Taking the Training Wheels Off: Aligning LLMs without Personas — LessWrong
• If you told an AI Alignment researcher in 2018 about an alignment plan that involved collecting trajectory information of moral experts at scale…
Critique of current AI safety bug bounty programs — LessWrong
The potential value of AI safety bug bounty programs Generally, AI labs should (and most do) put their models under extensive safety testing before d…
Can LLMs even teach? Exploring the Teacher Axis — LessWrong
TLDR As a passionate teacher, it has pained my heart to watch my students lose deeper critical thinking skills and independent reasoning. But attempt…
Systems Dynamics Model for Pausing AI — LessWrong
Summary This post documents our process of applying systems dynamics modeling to the problem of AI governance, tracing the feedback loops connecting…
Companions aren't Coaches: AIs' Effect on Social Skills — LessWrong
I’m wary of AI companions. By “AI companions”, I’m referring to conversational programs that are intended, either by the developer or the human user,…
"Contagious Humming" to Silence a Room — LessWrong
Often when running meetups you’ll have several lively conversations going at the same time. This is a great problem to have, but it can make it diffi…
Qualcomm’s seemingly backhanded compliment to Nvidia RTX Spark announcement
Nvidia's RTX Spark SoCs hope to bring new life to the Windows on ARM platform. These chips are designed for high-performance in slim and light laptops, with a l...
Dissolving the Deep Learning Sample Efficiency Gap — LessWrong
A common observation about deep learning is that it's wildly sample inefficient compared to humans. Deep learning systems appear to need much more re…
We Need Breadth-First AI Safety Plans — LessWrong
Depth-first plans lay out a path from here to aligned superintelligent AI. We need those kinds of plans. But depth-first plans depend on many assumpt…
Superintelligence of the gaps — LessWrong
Many classic AI doom scenarios rely on superintelligence using its vastly superior intelligence to outplan, outcompete and outkill you. …
Lean, not backpressure — LessWrong
Lucas Costa has written a good article on how to build systems that can handle code-generating robots. Unfortunately, when calling it backpressure, h…
My reactions to “I underestimated AI capabilities (again)” — LessWrong
An application response I wrote! Please feel free to leave any feedback! • …
“This Hypothetical is Unrealistic” is not a Valid Objection — LessWrong
Whenever a discussion touches ethics, philosophy, or relates to guiding principles, hypotheticals become useful. We cannot investigate every idea wit…
NLA Thought Anchors — LessWrong
The following post seeks to look further into why NLA (Natural Language Autoencoders) contains the prediction more often when the original activation…
Barriers to a Prosperous Future — LessWrong
The current race towards producing general artificial intelligence systems brings with it severe risks, yet no AI company developing frontier models…
The main impact from automated AI production: concentration of power? — LessWrong
There’s a lot of talk about automated AI R&D and the like. It’s been discussed since at least 1965 when statistician I.J. Good coined the term ‘intel…
Financial Costs of an AI Pause? — LessWrong
I’ve analyzed the near-term economic effects of an AI pause, out of concern for my investments, and a desire to predict how strong political oppositi…
Agriculture needs another revolution — LessWrong
Summary Vertical farming has the potential to unlock multiplicative yield gains per area of land and catalyze development of new technologies (precis…
Outrunning your headlights — LessWrong
> This is exactly the right place to probe. Gromov-Wasserstein is genuinely dimension free. Partial and semi-relaxed are precisely the mechanisms for…
05 Part 2 — LessWrong
Preface * ^means articles I read in full, otherwise assume I skimmed it * I show my discovery graph in (via …) blocks, those without (via …) usuall…
Ensemble monitoring for AI control: diverse signals outweigh more compute — LessWrong
tl;dr What improves LLM monitoring besides leveraging more compute? Leveraging more diverse compute. …
An attempted synthesis on probabilities and infinities — LessWrong
(This is the last post in my sequence. Reading the previous post on Infinite ethics and UDASSA is necessary for understanding this post. Reading the…
How to treat the AI consciousness topic with a more proper philosophy and science — LessWrong
I think that, IMO, ideally, it's best, that one treats AI consciousness topic with proper philosophy and science. It's IMO best, if anyone, on any
AI as Biology's Digital Microscope — LessWrong
History's greatest scientists have always used cutting-edge tools to explore the microscopic world around us. Antonie van Leeuwenhoek, with his micro…