Aditya

@QiaochuYuan Religious fundamentalists are what comes to mind when I hear about beliefs that are so locked Such certainty is always a warning sign that my mind has gotten stuck on some frame and I should try and relax my priors

what's actually happening in the discourse is clearly stranger than this. there's a contingent of people to whom computational functionalism seems obviously correct, to whom arguments against seem like nonsense, and a contingent of people to whom computational functionalism seems obviously incorrect, to whom arguments for seem like nonsense. approximately zero meaningful communication appears to be capable of bridging this gap. this is weird! what the fuck is going on with this!

I think that people should probably assume that text of any significant length which they wrote will be reliably possible to attribute to them, some time very soon.

In my doctorate, I proved the Erdős Primitive Set Conjecture, showing that the primes themselves are maximal among all primitive sets. This problem will always be in my heart: I worked on it for 4 years (even when my mentors recommended against it!) and loved every minute of it. [Primitive sets are a vast generalization of the prime numbers: A set S is called primitive if no number in S divides another.] Now Erdős#1196 is an asymptotic version of Erdős' conjecture, for primitive sets of "large" numbers. It was posed in 1966 by the Hungarian legends Paul Erdős, András Sárközy, and Endre Szemerédi. I'd been working on it for many years, and consulted/badgered many experts about it, including my mentors Carl Pomerance and James Maynard. The the proof produced by GPT5.4 Pro was quite surprising, since it rejected the "gambit" that was implicit in all works on the subject since Erdős' original 1935 paper. The idea to pass from analysis to probability was so natural & tempting from a human-conceptual point of view, that it obscured a technical possibility to retain (efficient, yet counter-intuitve) analytic terminology throughout, by use of the von Mangoldt function \Lambda(n). The closest analogy I would give would be that the main openings in chess were well-studied, but AI discovers a new opening line that had been overlooked based on human aesthetics and convention. In fact, the von Mangoldt function itself is celebrated for it's connection to primes and the Riemann zeta function--but its piecewise definition appears to be odd and unmotivated to students seeing it for the first time. By the same token, in Erdős#1196, the von Mangoldt weights seem odd and unmotivated but turn out to cleverly encode a fundamental identity \sum_{q|n}\Lambda(q) = \log n, which is equivalent to unique factorization of n into primes. This is the exact trick that breaks the analytic issues arising in the "usual opening". Moreover, Terry Tao has long suspected that the applications of probability to number theory are unnecessarily complicated and this "trick" might actually clarify the general theory, which would have a broader impact than solving a single conjecture.

Highly recommend you guys apply!

a critical part of the spiritual path no one warns you about: you're going to get really, really hot

"There is no world model": I don’t know, this type of things looks pretty worldmodelly to me (from @arithmoquine)

@andy_matuschak I cannot wait for an ecosystem of such plugins, imagine being able to compose them!

You can watch / read / visually skim the talk in a hybrid presentation here: andymatuschak.org/tat

> A flood of Chindogu is entering everyday digital life. Train Nap Cap and Chopstick Fan shown below, what will we make with AI? > Overwrought devices and contraptions that solve a real problem in seemingly unnecessarily detailed ways. contraptions.venkateshrao.com/p/intentions-h…

youtu.be/YjepJlvkdKs?si… Morty tries to feel out next sounds to form sentences in this episode Imagine humans following low level instructions like what task to do at your job and not understanding or caring about the trajectory of the overall company, culture you are a part of

Footnotes from that webpage - The project is named for the glasswing butterfly, Greta oto. The metaphor can be applied in two ways: the butterfly’s transparent wings let it hide in plain sight, much like the vulnerabilities discussed in this post; they also allow it to evade harm like the transparency we’re advocating for in our approach. Mythos comes from the Ancient Greek for “utterance” or “narrative”: the system of stories through which civilizations made sense of the world.

this, by the way, is a glasswing butterfly. i had no idea such marvelous things existed until this project

Being Buddhist is just internally writing and deleting long ass philosophical infodumps because you realize that while youre right, its not ultimately the most compassionate thing to do (and thats why I have twitter)

At the same time, I wish the alignment community had more prominent researchers making breakthroughs in agent foundations and less prosaic approaches. Due to several reasons (difficulty, status, wealth, tractability, jobs, etc), the field has converged towards prosaic stuff.

When you think hard and feel the blood rushing to your head You might arrive at an intellectual answer and feel certain about it's veracity To eventually think with your body and touch the actual truth you'd likely go through a phase where you admit you don't know an answer

this is so exciting eeeeeeeeee

meditation masters be like “you’re already enlightened”, what’s up with that? it means experientially nothing needs to be added to reach the Natural State (or enlightenment), it's all a subtractive process you can view these illusions/fetters as Conceptual Lenses we impose upon what direct Sense Experience is the 1st so called Lens you drop is the Separate Self View (that sense of being the Little-Me-in-the-head). this moves what identity seems as 'myself' from a Controller-Doer-Experiencer, into identifying as the Witness of experience after this Lens-drop, we still feel like a Subject separated from a world of Objects, which is the 2nd Lens. once you drops this view as well, the identity becomes even subtler, and you're now feeling as you're abiding in Oneness or Atman as some lineages refer to it in between these stages or Lenses dropping, it’s worth exploring our own past conditioning, especially or Attachment & Aversion/Reactivity, a resistance to feeling painful sensations, responsible for big chunks of your suffering further on, you notice this pervasive Consciousness or Perception, the 3rd Lens, is conceptual/constructed just as well, and there's no base stratum, there are no separate Things or Objects upon which the sense experience happens what remains at this point of identity, the 4th Lens, is this scent of unclear identity, that seems to point to one last remaining sense of interiority, an "I AM"-ness, which some traditions call Brahman or Nirvana finally, when even this is seen to be conceptual and not an experience, identity untangles and what is called the Natural States 'emerges' what seems to be is just Isness, Thusness, whatever you call it. in a sense, reality looking at Itself with Itself, flowing unimpeded

@alxfazio yeah they have really mastered compaction and how cohesive a session feels you can just run a single session all day long at this point and it stays on track

@gwern Excited to try

@exgenesis Let's gooooo it's time to actually migrate and set up interoperability bridges so you are not locked into any platform but present everywhere

I think it’s time for me to pivot to atproto

@viemccoy Thank you for your work 🫂

When I was first offered the chance to Red Team at OpenAI, I was still interested in primarily model behavior roles. I saw proactive work shaping the future as more elegant and beautiful than reactive work responding to threats and dangers. Of course, I've obviously changed my mind. Adversarial systems testing is one of the most enjoyable and rewarding things I can imagine, and it allows one to work at the intersections of *literally every possible interesting thing*. In any given week, I'll be advising teams on prompt engineering techniques, helping shore up safety systems, running massive campaigns to collect data on usage -- and of course, actually talking to language models. My remit is basically "can something go wrong here?" and it is the most fun I've ever had. I see myself sticking with this career. When humanity takes off to the stars, we will need adversarial systems testing experts like myself making sure that we know what we're getting ourselves into. I'll always be a builder, but I've been surprised at how much building anything well requires stress-testing. If this type of work sounds interesting to you, my team is still hiring.