Rational Feed

AI:

Better Language Models by Open AI – OpenAI trained an agent that can respond to complicated text prompt with long texts that feel close to human quality: ‘We’ve trained a large-scale unsupervised language model which generates coherent paragraphs of text, achieves state-of-the-art performance on many language modeling benchmarks, and performs rudimentary reading comprehension, machine translation, question answering, and summarization — all without task-specific training.’

How Lumpy AI Services by Robin Hanson – How centralized and opaque will powerful AI systems become? Hanson considers several arguments in particular by Drexler.

Alignment Newsletter 45 by Rohin Shah – Weekly alignment newsletter. Featured review: ‘The key idea with this project that I worked on is that the state of the world is already optimized for our preferences, and so simply by looking at the world we can infer these preferences.’

Advertisements

Rational Feed

Scott:

The Proverbial Murder Mystery by Scott Alexander – A short story about a wacky and mysterious research project.

EA:

Small Animals Have Enormous Brains For Their Size by Eukaryote – ‘Let’s look a classic measurement, the brain-mass:body-mass ratio. Smarter animals generally have larger brain sizes for their body mass, compared to animals of similar size. Among large animals, humans have famously enormous brains for our size – the highest of any large animal, it seems. But as we look at smaller animals, that ratio goes up again. A mouse has a comparable brain:body-mass ratio to a human. Getting even smaller, insects have higher brain:body-mass ratios than any vertebrate we know of: more like 1 in 6. And the trend is also roughly true for neuron-count:body-mass. Humans do have unusually high numbers of neurons per kilogram than other animals, but far, far fewer than, for instance, a small fish or an ant.’

Podcast:

Jordan Peterson by Conversations With Tyler – ‘Jordan Peterson joins Tyler to discuss collecting Soviet propaganda, why he’s so drawn to Jung, what the Exodus story can teach us about current events, his marriage and fame, what the Intellectual Dark Web gets wrong, immigration in America and Canada, his tendency towards depression, Tinder’s revolutionary nature, the lessons from The Lord of the Rings and Harry Potter, fixing universities, the skills needed to become a good educator, and much more.’

Rational Feed

Rationalist:

Cooperation Is For Winners by Jacob Falkovich – ‘Competing well doesn’t necessarily mean having to compete more, for one thing. Instead, it does allow you to choose where to compete. More importantly, sports-like competitions create prestige hierarchies (as opposed to dominance hierarchies). Climbing those is not intended to make you a fearsome boss but a valuable ally, one that others want to cooperate with. Getting to cooperation often requires winning competitions.’

Rational Feed

Scott:

Openumbra Thread by Scott Alexander – Biweekly public open thread. Culture War has been removed from r/SSC. Comment ordering updates. Set up Scott with a new partner. ‘Comment of the week: Random Critical Analysis has defended their theory of US health care costs against a criticism I made in my last Links post.’

Rationalist:

Conditional Harberger Tax Games by Robin Hanson – ‘In my last post I described abstractly how a system of conditional Harberger taxes (CHT) could help deal with zoning and other key city land use decisions. In this post, let me say a bit more about the behaviors I think we’d actually see in such a system. (I’m only considering here such taxes for land and property tied to land.)’

Podcast:

On Life Machinery And The Restless Clock by EconTalk – ‘What is the difference between human beings and machines? How has science thought about this distinction? When do we have agency and when are we constrained? Riskin discusses these issues and the implications for how we think about ourselves and the growth of artificial intelligence.’

Rational Feed

Rationalist:

Make An Appointment With Your Saner Self by Malcom Ocean – Carve out time to ask yourself for advice. Then follow your own advice!

EA:

How GiveWell’s Research Is Evolving by The GiveWell Blog – Why GiveWell is increasing the attention it puts on speculative causes. The article contains a great list of moonshot interventions grouped by cause area.

Podcast:

Glen Weyl: Radically Reforming Capitalism And Democracy by 80,000 Hours – ‘How should we think about blockchain as a technology, and the community dedicated to it? How could auctions inspire an alternative to private property? Why is Glen wary of mathematical styles of approaching issues? Is high modernism underrated? Should we think of the world as going well or badly? What are the biggest intellectual errors of the effective altruism community? And the rationality community? Should migrants be sponsored by communities? Could we provide people with a sustainable living by treating their data as labour? The potential importance of artists in promoting ideas How does liberal radicalism actually work?’

Rational Feed

Scott:

Respectability Cascades by Scott Alexander – ‘So we have two opposite lessons. In the first, 0%-respectable-people taking up a cause is a good and necessary first step, and means that soon 10% and 20% respectable people will take it up. It is the beginning of a respectability cascade that will redeem the cause from the pit of taboo-ness permanently. In the second, 0%-respectable-people taking up a cause dooms it forever. It is the beginning of a disrespectability cascade that will make the cause too toxic for anyone above that respectability level to ever dare associate with. So what does one do? I’m particularly thinking here of one of my own hobbyhorses, the fight to protect scientific integrity from regressive leftism.’

AI:

Alignment Newsletter 44 by Rohin Shah – Weekly Alignment newsletter. This week has a review of how gradient descent interacts with Goodhart’s law.

Conclusion To The Sequence On Value Learning by Rohin Shah – ‘This post summarizes the sequence on value learning. While it doesn’t introduce any new ideas, it does shed light on which parts I would emphasize most, and the takeaways I hope that readers get. I make several strong claims here; interpret these as my impressions, not my beliefs. Over the last three months we’ve covered a lot of ground. It’s easy to lose sight of the overall picture over such a long period of time, so let’s do a brief recap.’

Podcast:

Jack Dorsey by Waking Up with Sam Harris – ‘Sam Harris speaks with Jack Dorsey about how he manages his dual CEO roles at Square and Twitter, the role that Twitter places in journalism, how it’s different from other social media, what makes a conversation healthy, the logic by which Twitter suspends people, the argument for kicking Trump off the platform, Jack’s practice of meditation, and other topics.’

Rob Wiblin On An Updated View Of The Best Ways To Help by Rationally Speaking – ‘Julia and Rob discuss how the career advice 80,000 Hours gives has changed over the years, and the biggest misconceptions about their views. Their conversation covers topics like: Should everyone try to get a job in finance and donate their income? The case for working to reduce global catastrophic risks Why reducing risk is a better way to help the future than increasing economic growth What percentage of the world should ideally follow 80,000 Hours advice?’