Binit Maharjan
3 followers · 5 following · 190 views
on the atlas — 25
- Let's talk about the AI coordination problem — LessWrong1 savers
- James M. Buchanan Jr. – Prize Lecture - NobelPrize.org1 savers
- Where’s my ten minute AGI? - by Anson Ho - Epoch AI1 savers
- Radical Optionality — Governing Transformative AI Under Uncertainty6 savers
- On Private Governance - by Dean W. Ball - Hyperdimensional2 savers
- The most overlooked roles in AI safety | 80,000 Hours2 savers
- Plans A, B, C, and D for misalignment risk — LessWrong6 savers
- How to be More Agentic - by Cate Hall - Useful Fictions24 savers
- Taking Jaggedness Seriously - by Helen Toner - Rising Tide1 savers
- A global workspace in language models \ Anthropic25 savers
- The "Messy Middle" - by Molly Kinder3 savers
- If AI is normal technology, history is not reassuring. — LessWrong2 savers
- A crash course on US air pollution - Andy Masley1 savers
- Top Performers are Pathologically Ambitious - by Matt Beard3 savers
- How to Buy Cheap Claude Tokens in China - by Zilan Qian8 savers
- AI policy must fail gracefully - by Nat Purser1 savers
- entropy.pdf1 savers
- What happens when AI starts building itself? | TechCrunch2 savers
- Man-Computer Symbiosis5 savers
- Curius / Onboarding2621 savers
- Why books don't work64 savers
- The third wave of American philanthropy - by Nan Ransohoff17 savers
- How Go Players Disempower Themselves to AI — LessWrong9 savers
- My hobby: running deranged surveys — LessWrong3 savers
- The Post-COVID Decline in the Labor Share - Liberty Street Economics2 savers
highlights — 45
When we told it not to think about something, the concept lit up in its J-space less than when we said it should think about it, but much more than when we never mentioned it
A global workspace in language models \ AnthropicBut when one of these patterns lights up, it doesn’t mean the model is saying that word—just that the word is on its mind. If you've heard of language models having a "scratchpad" or “chain of thought”—text they write to themselves while reasoning—the J-space is something different
A global workspace in language models \ AnthropicThere is no version of universal income transfer that solves this without enormous wage inflation in essential sectors, which has its own cascading consequences
The "Messy Middle" - by Molly Kinderwhy would anyone believe the political will appears once it becomes a fact, in a country whose institutions are weaker, more polarized, and more captured by concentrated wealth than they were the last time we built the welfare state
The "Messy Middle" - by Molly KinderIf AI is a general-purpose technology that erodes the cognitive premium broadly, then the destinations of similar pay/ skills are shrinking at the same time as the displaced are searching for them. It becomes a game of musical chairs with fewer chairs each round, at least among the knowledge jobs.
The "Messy Middle" - by Molly KinderPlumbers are well paid in part because of scarcity
The "Messy Middle" - by Molly KinderThere is no geographic or cultural distance between the displaced and the powerful
The "Messy Middle" - by Molly Kindermany other classes of transformative change end up disruptive in ways that hurt before they can help
If AI is normal technology, history is not reassuring. — LessWrongOf course, the longer term impact was positive; those benefits are not in question[2] - not that those alive during the transition should have cared.
If AI is normal technology, history is not reassuring. — LessWrongthe main way emissions create additional PM2.5 and ozone going forward might be via their CO2.
A crash course on US air pollution - Andy MasleyWithin-urban disparities often reflect the legacies of redlining.
A crash course on US air pollution - Andy Masleyfarmers are politically untouchable in America
A crash course on US air pollution - Andy MasleyAgriculture causes about 12% of total US anthropogenic PM2.5 mortality, and the large majority of that is from livestock
A crash course on US air pollution - Andy MasleyGas-fired power plants are now the largest stationary NOx source in America, and this is being driven way up by the data center buildout.
A crash course on US air pollution - Andy MasleyThe Clean Air Act addresses this through what EPA calls the “Good Neighbor” provision, which requires every state’s SIP to prohibit emissions that significantly contribute to another state’s nonattainment, or interfere with another state’s ability to maintain the standards.
A crash course on US air pollution - Andy MasleyYou might think their decision to set this level has something to do with balancing health against the economic costs of restricting air pollution, but it is actually illegal for them to consider economic costs when setting the clean air threshold, according to a unanimous Supreme Court case in 2001.
A crash course on US air pollution - Andy MasleyWhite Americans on average are exposed to about 17% less PM2.5 than they cause through consumption, while Black Americans are exposed to about 56% more and Hispanic Americans about 63% more than they cause.
A crash course on US air pollution - Andy MasleyThe EPA knows this is a tension, and there are different reasons for why they use the model they do, but one of them is political. In 2003, EPA tried using an age-adjusted value in a Clear Skies analysis where it ranked the lives of people over 70 as having a slightly lower dollar value. Critics in government and the media called it a “senior death discount,” and EPA dropped it.
A crash course on US air pollution - Andy MasleyBut these paint a very different picture than “10,000 cars driving thousands of miles.”
A crash course on US air pollution - Andy MasleyA two-stroke leaf blower has minimal emission controls, it’s basically a small, dirty engine with a fan attached.
A crash course on US air pollution - Andy MasleyA new truck emits roughly 99% less hydrocarbons per mile than a 1970 pickup truck.25
A crash course on US air pollution - Andy Masleyroughly 70,000 deaths PM2.5 causes
A crash course on US air pollution - Andy Masleyair pollution carries some level of harm at any level
A crash course on US air pollution - Andy MasleyAir pollution in America kills more people than car accidents, but most of the people who die due to long-term exposure to air pollution are much older, and often lose a few years of expected life.
A crash course on US air pollution - Andy MasleyOne study found that people who didn’t smoke but used wood stoves for at least 30 days per year were twice as likely to get lung cancer as non-smokers who didn’t use the stoves
A crash course on US air pollution - Andy Masleyit’s just as important to ask: “greatness at what?”
Top Performers are Pathologically Ambitious - by Matt BeardJensen berates his biographer for asking too many questions about AI safety
Top Performers are Pathologically Ambitious - by Matt BeardIt was just time for Jensen to yell at his stove
Top Performers are Pathologically Ambitious - by Matt Beardyou’ve basically never worked hard.
Top Performers are Pathologically Ambitious - by Matt Beardif you genuinely care about improving the world, you should aim much higher and work much harder than the average person
Top Performers are Pathologically Ambitious - by Matt Beardpotential promotion, fraud, and even blackmail based on leaked users’ data from the proxy
How to Buy Cheap Claude Tokens in China - by Zilan Qianthe log harvest is the actual margin
How to Buy Cheap Claude Tokens in China - by Zilan Qianideal dataset for post-training
How to Buy Cheap Claude Tokens in China - by Zilan QianThe logs are the product
How to Buy Cheap Claude Tokens in China - by Zilan QianThe Worldcoin black market offered a documented precedent, with iris scans harvested from KYC merchants in Cambodia and Kenya, sold for under $30.
How to Buy Cheap Claude Tokens in China - by Zilan QianSingapore, with a population smaller than that of New York City, “surprisingly” leads global per capita use of Anthropic’s Claude in April 2026.
How to Buy Cheap Claude Tokens in China - by Zilan Qiancapacity housed in statute is better positioned to endure changes in administration than capacity distributed across agencies.
AI policy must fail gracefully - by Nat PurserState attorneys general presently serve on the front lines of AI enforcement, but most lack the technical staff to evaluate the products being deployed within their borders
AI policy must fail gracefully - by Nat PurserRisks from AI and threats to democracy compound each other
AI policy must fail gracefully - by Nat Purserone can hardly take a military commander or a corporation president away from his work to teach him to type
Man-Computer Symbiosis. The computer should read the man's writing
Man-Computer SymbiosisMen naturally speak redundant languages organized around unitary objects and coherent actions and employing 20 to 60 elementary symbols. Computers "naturally" speak nonredundant languages, usually with only two elementary symbols and no inherent appreciation either of unitary objects or of coherent actions.
Man-Computer SymbiosisThey would be easier to solve, and they could be solved faster, through an intuitively guided trial-and-error procedure in which the computer cooperated, turning up flaws in the reasoning or revealing unexpected turns in the solution.
Man-Computer Symbiosisthose years should be intellectually the most creative and exciting in the history of mankind.
Man-Computer Symbiosisenable men and computers to cooperate in making decisions and controlling complex situations without inflexible dependence on predetermined programs
Man-Computer Symbiosis