Intelligence is compression. E.g. instead of memorizing every single equation in a certain mathematical topic, someone that truly understands the subject, can usually derive these equations. In other words, they are able to compress knowledge of the field very effectively.
I think the ml models we currently have are terrible at compression. Thats why they are hundreds of gigabytes large, while still making embarassing mistakes. Most of the benefits we get from scaling is simply immitating intelligence better. Therefore, performance over size (in GB) is a more appropriate metric for comparing how smart models are.
Prediction: Quantum search algorithms, such as Grovers algorithm will massivel accelerate ML tasks by allowing tractable search for functions that describe model weights more effectively.
After photolithography, semiconductor manufacturing will likely turn to molecular manufacturing - that is, growing processors just like crystals. This is a lot less far-fetched than it may sound. In order for this to work though, we would need a highly symmetrical processor design. The reason for this is that crystal growth creates symmetrical, repeating structures. Interestingly, the internal structure of basic FPGAs, are also highly symmetrical (logic/memory cells embedded in a routing fabric). Differentiation of the compute-crystal would happen at the "gateware" level. The difficult part is really to find a building block (unit cell) that we can grow such a crystal from. Of course, these building blocks would be many orders of magnitude smaller than those of an FPGA. Conceivablby, a building block could span millions of atoms and effectively perform something akin to NAND+hysteresis+routing.
Nov 24 2025
i grew up in a household where the legitimacy of basic human rights were completely nonnegotiable. it would have been so utterly out of line for anyone to suggest that a human may not deserve food, or that someone deserves death, that it never occurred to me that these are things that discussions can be had about.
recently i have found myself confronted with the opposite. i am seemingly surrounded by people who regularly argue the exact other way around. to them it seems to be completely logical that some people don't deserve to live. this experience is completely disorienting and my foundations of belief are shaken. my understanding about where the axioms of reasoning lie, are apparently completely incorrect. as a matter of fact, it almost seems like there are no axioms that we can agree on. this is catastrophic, because as long as we all agree that human suffering is a bad thing, even if we cant find common ground on *how* we should go about reducing it, we can still agree on the goal. this does no longer seem to be the case. things have become debatable that previously were completely out of bounds for discussion. if a misanthrope poses the question "should a murderer die", then answering to that question even if the answer is no, is in itself already acknowledging that the question of basic human rights is something that can be debated about. even worse, you may be deep into a dialog, trying to explain basic empathy, but as soon an oblivious third person joins in, and backs up the first person, even if its just in one point that is tangent to the discussion, then you have lost and the misanthropy is suddenly legitimized. this is catastrophic, and as far as i can tell entangling yourself into discussions such as these shall be avoided, they really just serve to strengthen anti-human beliefs. argue on a higher level, question the question.
i am still unsure on how to parse the bigger picture of this apparent loss in empathy. maybe i truly am naive and i just have to "come to my senses" and realize that "the world is evil and that's just how things are", but this seems completely unreasonable to me. any person that believes in human agency should be able to see that we are involved in taking the current situation and turning it into the next. it could be that the current situation is bad, that maybe we even seem to be going further into the wrong direction, however it is *astoundingly stupid* to say, "that's just how it is" and not do anything about it. in other words - we have the ability to take action to turn a bad situation into a good one, it is all about the decisions we make, and it is important that the system of belief which we base them on is one that will lead to a better future, instead of one completely driven by primal emotions.
however, some people simply do not care, and even by my standards, saying that "there are no evil people and that there is a reason for every villain to be the way they are" would be naive. what i mean is - not every antagonist was beaten as a child. i think there are people who simply do not have a strong sense of empathy, maybe this is literally caused genetically, maybe by societal factors, maybe this is just caused by indulgence in non-determinism, which allows people to simply follow their emotions like a broken compass and completely evade moral responsibility for their actions. we can't do much about people with a broken compass but i think it is easy to see - there is no reason for someone to play a game if they don't care for it. we just need enough players who care, and systems to constrain those that don't.
it should easily be possible to find out if there is a per-person correlation between blasphemous actions and "godly punishments" e.g. cancer, accidents, tragedy, etc. actually, we could go even further and, considering the rates of blasphemy correlated to the rate of punishment, find out which religion god likes most
Feb 19 2025it's possible that, had chernobyl and fukushima never happened, and had the public opinion of nuclear energy therefore not been skewed, we would likely not be discussing climate change today.
Feb 18 2025i propose that almost all problems that the world is facing today can be solved by advancing scientific research. i also propose that the fields of science, whether advanced through academia or industry, are interconnected and rely on one another. e.g. medicine as a whole and pharmacology in specific are interlinked and progress in one will lead to progress on the other. some fields have more unidirectional relationships - e.g. computer science is more or less upstream of aviation (enables things like cfd), and lithography is upstream of computer science. some fields are upstream of many other fields. i believe that in order to most effectively progress human development, we should put more effort into pushing the fields that are furthest upstream. in a simplified world, this is some sort of flow optimization problem on a directed graph, where the weights correspond to the dependencies of research areas on one another.