Blake Elias
17 followers · 4 following · 771 views
on the atlas — 87
- The Diamond Age1 savers
- Hysteresis5 savers
- Overton window5 savers
- Liquidity in Drug Discovery – Bharath Ramsundar – Decentralized Protocol and Deep Drug Discovery Researcher1 savers
- [1902.09469] Embedded Agency1 savers
- View article22 savers
- All Possible Views About Humanity's Future Are Wild6 savers
- View article3 savers
- A beautiful game: harnessing the value of superior Intelligence1 savers
- Dr. AI or: How I Learned to Stop Worrying and Love Economics · Issue 1.1, Summer 20191 savers
- The Frame-Dependent Mind3 savers
- A graduate school survival guide: "So long, and thanks for the Ph.D!"3 savers
- Quantum Gravity (Stanford Encyclopedia of Philosophy)1 savers
- The Philosopher-Builder - Cosmos Institute7 savers
- Full-Stack Alignment: Co-Aligning AI and Institutions with Thick Models of Value | Research Paper2 savers
- you can just do stuff - alice maz9 savers
- The Only Important Technology Is The Internet — Kevin Lu17 savers
- Sovereign Intelligence1 savers
- I Don't See How Comparative Advantage Applies In a World of Strong AI1 savers
- Grounding the Conversation About AI - by Steve Newman1 savers
- The Fun Theory Sequence - LessWrong2 savers
- AI 20273 savers
- Societal and technological progress as sewing an ever-growing, ever-changing, patchy, and polychrome quilt — LessWrong1 savers
- Acute Risk Period - LessWrong1 savers
- musings and rough drafts – unpolished, semi-edited, streams of thought2 savers
- Can AI Scaling Continue Through 2030? | Epoch AI6 savers
- The Project Gutenberg eBook of On Liberty, by John Stuart Mill.1 savers
- Be less scared of overconfidence - LessWrong2 savers
- Why You Should Start a Blog Right Now39 savers
- Why and how to write things on the Internet — LessWrong1 savers
- Staring into the abyss as a core life skill — LessWrong7 savers
- Anomie2 savers
- Why AI Will Save the World | Andreessen Horowitz2 savers
- AI 20279 savers
- Navigating an AGI Transition - Google Docs1 savers
- Non-Human Words: On GPT-3 as a Philosophical Laboratory | American Academy of Arts and Sciences1 savers
- What 2026 looks like — AI Alignment Forum4 savers
- Why Tool AIs Want to Be Agent AIs · Gwern.net22 savers
- Is OpenAI OK?, by Gwern · Gwern.net1 savers
- Autopoiesis4 savers
- Forgetting to forget | Joel Gustafson2 savers
- An Intuitive Explanation of Solomonoff Induction - LessWrong8 savers
- 1. Design and Science · Journal of Design and Science2 savers
- The Gervais Principle, Or The Office According to “The Office”12 savers
- Wanted: Smart Generalists2 savers
- Maslow's hierarchy of needs4 savers
- Flat Affect, or Emotional Indifference - Exploring your mind3 savers
- On Being an Illegible Person13 savers
- Putting Ideas into Words12 savers
- The Gervais Principle III: The Curse of Development3 savers
- Magna Carta Scientiae5 savers
- Why to Start a Startup in a Bad Economy5 savers
- It's Time to Govern - Letters to a Young Technologist7 savers
- The Equity Equation3 savers
- Going From California With An Aching In My Heart. | Slate Star Codex12 savers
- Beyond Smart7 savers
- #6: Becoming Hyper-Local Real Estate Guy2 savers
- Printable2 savers
- Curius / Onboarding2667 savers
- Fast · Patrick Collison48 savers
- How to Work Hard29 savers
- Research as Understanding26 savers
- The myth of the myth of the well read person20 savers
- How to Pick Your Life Partner - Part 2 — Wait But Why18 savers
- The Marginalian – Marginalia on our search for meaning.16 savers
- Goodhart's law15 savers
- A Vision of Metascience13 savers
- How to Drop Out11 savers
- Kernel Magazine | Reclaiming the Future11 savers
- Interact10 savers
- Google Search Is Dying | DKB10 savers
- Early Work10 savers
- Nobody Is Perfect, Everything Is Commensurable | Slate Star Codex10 savers
- The web browser as a tool of thought | thesephist.com9 savers
- Project Ideas – Future Fund7 savers
- The Log: What every software engineer should know about real-time data's unifying abstraction | LinkedIn Engineering7 savers
- Don't try to cure cancer — Ada Nguyen7 savers
- How much do you believe your results? - LessWrong6 savers
- Notes on technology in the 2020s6 savers
- Open-endedness: The last grand challenge you’ve never heard of – O’Reilly5 savers
- A Declaration of the Interdependence of Cyberspace4 savers
- Thou Art Godshatter - LessWrong4 savers
- Embedded Agency (full-text version) - LessWrong3 savers
- Google Calendar - Week of January 3, 20223 savers
- After Virtue3 savers
- Evolutionary Game Theory: Theoretical Concepts and Applications to Microbial Communities2 savers
- Category Theory on Math3ma2 savers
highlights — 1752
There could be rewards for this work too.
A beautiful game: harnessing the value of superior Intelligencea period equal to the time in which the output has been validated to be scrutable and comprehended by at least one Normal intelligent being, e.g. adult human with IQ ≤ 130.
A beautiful game: harnessing the value of superior Intelligencea maximally accurate (e.g. true) representation of what manifests in physical reality
A beautiful game: harnessing the value of superior IntelligenceTo prevent this, an ISR should first be routed to this network to look for an existing solution
A beautiful game: harnessing the value of superior Intelligenceefficient and credible punishment of anti-social behavior via: (a) negative reputation scoring,
A beautiful game: harnessing the value of superior Intelligenceewards distributed as before
A beautiful game: harnessing the value of superior Intelligencerewards distributed as before.
A beautiful game: harnessing the value of superior Intelligenceand rewards distributed as before
A beautiful game: harnessing the value of superior Intelligence(b) requires much lower work than the provision (a) of an Intelligence Service Output, this is such that the verification is inexpensive compared to provision of the service hence ensuring the output is maximally correct and aligned to the customer i.e. minimizing agency costs.
A beautiful game: harnessing the value of superior Intelligenceoften associated directly and immediately with the scientific revolution,
A beautiful game: harnessing the value of superior IntelligenceConsider the limit condition, where negative value information (e.g. lies) is constantly used on entire societies, the consequences are "not that you believe the lies, but rather that nobody believes anything any longer. […] lies, by their very nature, have to be changed. […] And a people that no longer can believe anything cannot make up its mind. It is deprived not only of its capacity to act but also of its capacity to think and to judge. And with such a people you can then do what you please”. Hannah Arendt (1978).
A beautiful game: harnessing the value of superior Intelligencewho complains that he wishes philosophers would not restrict themselves to “commenting and polishing the present fragmentary physical theories, but would take the risk of trying to look ahead” (Rovelli, 1997, p. 182).
Quantum Gravity (Stanford Encyclopedia of Philosophy)thermodynamics as destiny
The Philosopher-Builder - Cosmos InstituteThey defer to easy answers: what customers seem to want, the tribal consensus reinforced by algorithmic feeds, the technical frameworks they unconsciously expand into worldviews.
The Philosopher-Builder - Cosmos InstituteEvery builder’s first duty is philosophical: to decide what they should build for. This duty has largely been forgotten.
The Philosopher-Builder - Cosmos InstituteSystems capable of extrapolating value-aligned responses to new situations at AI speed while preserving democratic principles.
Full-Stack Alignment: Co-Aligning AI and Institutions with Thick Models of Value | Research PaperAt first glance, many researchers might find it weird (or a distraction) that in order to make strides in research, we should turn to product. But actually I think it is quite natural: assuming that we care that AGI does something beneficial for humans, and not just act intelligent in a vacuum (as AlphaZero does), then it makes sense to think about the form factor (product) that AGI takes on – and I think the co-design between research (pretraining) and product (internet) is beautiful.
The Only Important Technology Is The Internet — Kevin LuWhy you should care about product-research co-design, and what is the dual of RL?
The Only Important Technology Is The Internet — Kevin Luthe proper role for product, as it relates to pure research, is to sift through the material the researchers produce, looking for rough diamonds to polish and bring to market
you can just do stuff - alice mazlicklider and taylor were men
you can just do stuff - alice mazI think our institutions have failed, and it falls to the unaffiliated to push us forward again, people acting toward their own ends
you can just do stuff - alice mazMuch like RL agents or video-game players want to win (and therefore discover new strategies), online users want to contribute new ideas (sometimes receiving upvotes or ad revenue), hence expanding the frontier of knowledge and creating a natural learning curriculum.
The Only Important Technology Is The Internet — Kevin Lugenerally we might consider textbooks to represent the pinnacle of human intelligence
The Only Important Technology Is The Internet — Kevin Lu(Apparently, we have not made great progress on new paradigms :)
The Only Important Technology Is The Internet — Kevin LuIf you don’t control your AI stack, you surrender your fate to those who do. Not our weights, not our society.
Sovereign IntelligenceThose who control AI
Sovereign Intelligencewhether it serves the purposes for which you’ve adopted it.
The Frame-Dependent Minderefore so is morality itself frame-dependent.
The Frame-Dependent Mindand also as parts of a greater collective whole at the same time, without surrendering that individuality.
The Frame-Dependent Mindsocieties seem not to allow this.
I Don't See How Comparative Advantage Applies In a World of Strong AILower quality work. In the world we’re talking about, AIs are likely “better” than humans not only in being able to do things faster, but also at higher quality and more reliably. I’m not likely to want to be operated on by a human surgeon, when an AI surgeon would have a higher success rate. We’re unlikely to put a person behind the wheel of a delivery van, if AIs have become much safer drivers. In general, we’re unlikely to put the inputs for any given job into the hands of a human, when an AI can give us a better product from those same inputs.
I Don't See How Comparative Advantage Applies In a World of Strong AITransactional overhead. For many jobs, an AI will be able to get the job done in less time than it would take me merely to explain what I want to a person. Not to mention all of the overhead of finding the right person, agreeing on a price, and so forth.
I Don't See How Comparative Advantage Applies In a World of Strong AIAdaptation. Even if there are some jobs at which a human could find gainful employment, the nature of those jobs might change too quickly for people to keep up. (Smith notes this, but somehow doesn’t allow it to influence his primary conclusion.) In a world in which we have universal AI and it’s so valuable that we’re building more AIs as fast as we possibly can, rapid change seems likely.
I Don't See How Comparative Advantage Applies In a World of Strong AIThere’s a reason this doesn’t currently exist: it will require a lot of effort – for moderation, analysis, and curation. I have thoughts for how to achieve this at scale, but that’s a topic for another day. We’ll start small.
Grounding the Conversation About AI - by Steve Newmanunlike those sites, we’ll actively recruit panels on topics of importance, and facilitate discussions that explore all of the nooks and crannies. We’ll maintain accessible summaries of each discussion, rather than expecting readers to dive into enormous discussion threads. And we’ll continuously update each topic over time.
Grounding the Conversation About AI - by Steve NewmanOn the potential for nipping the next pandemic in the bud through strategic monitoring of viruses in wastewater:
Grounding the Conversation About AI - by Steve NewmanI disagree with that paper -- I don't think you could meaningfully separate those "concerning" types of experiments from the rest of virology. Also, the scientific community and NIH have all had this discussion already, and they identified a set of research (GOF research) as the type of research that might require additional oversight. It's a sensitive topic and [redacted] is outside the mainstream on this point.
Grounding the Conversation About AI - by Steve NewmanWith a day or two of work (spread over a longer calendar period), I accomplished more than I might have with months of solitary research.
Grounding the Conversation About AI - by Steve NewmanThese were topics I didn’t know much about. I started out by plowing through a bunch of papers and reports… and ran into all of the problems mentioned above: apparent contradictions, implausible claims, difficulty in understanding which facts were important.
Grounding the Conversation About AI - by Steve Newmanpossible to assemble shorter DNA fragments into these longer lengths, but the process requires skill and expertise and is not amenable to automation.
Grounding the Conversation About AI - by Steve NewmanSomehow no one gets around to answering these big-picture real-world questions: what are the key levers that determine the equilibrium in cybersecurity, or misinformation in politics, or the productivity of software engineers. This might explain why discussion of AI risks and benefits is often out of touch with the real world, leading to acrimony and bad takes.
Grounding the Conversation About AI - by Steve NewmanThe rate at which value can increase as more bits are added, and the rate at which value must increase for eudaimonia, together determine the lifespan of a mind.
The Fun Theory Sequence - LessWrongour present world is clearly not the result of such optimization
The Fun Theory Sequence - LessWrongMany critics (including George Orwell) have commented on the inability of authors to imagine Utopias where anyone would actually want to live.
The Fun Theory Sequence - LessWrongWe pragmatists think of moral progress as more like sewing together a very large, elaborate, polychrome quilt, than like getting a clearer vision of something true and deep." – Richard Rorty (2021) pg. 141
Societal and technological progress as sewing an ever-growing, ever-changing, patchy, and polychrome quilt — LessWrongthe period of human history wherein it is possible to destroy all of civilization.
Acute Risk Period - LessWrong“you shouldn’t work for an organization unless you’re at least a little bit trying to reform it”. That stuck with me, and changed my concept of “working for an org”.
musings and rough drafts – unpolished, semi-edited, streams of thoughtThere are examples of efficient encoding of images with as few as 32 tokens, which after being adjusted by typical text dictionary size would result in 22 tokens per image.85 We take 22 tokens per image and second of video as a central guess,
Can AI Scaling Continue Through 2030? | Epoch AIWe will argue that multimodal data will result in effective data stocks of 450 trillion to 23 quadrillion tokens, allowing for training runs of 6e28 to 2e32 FLOP. Furthermore, synthetic data might enable scaling much beyond this if AI labs spend a significant fraction of their compute budgets on data generation.78
Can AI Scaling Continue Through 2030? | Epoch AIBut the peculiar evil of silencing the expression of an opinion is, that it is robbing the human race; posterity as well as the existing generation; those who dissent from the opinion, still more than those who hold it. If the opinion is right, they are deprived of the opportunity of exchanging error for truth: if wrong, they lose, what is almost as great a benefit, the clearer perception and livelier impression of truth, produced by its collision with error.
The Project Gutenberg eBook of On Liberty, by John Stuart Mill.