4 ms·

I think part of the issue is that a lot of the 'big' advances in physics nowadays aren't constrained by brain power as much as they are by cost (or rather, mone

I think part of the issue is that a lot of the 'big' advances in physics nowadays aren't constrained by brain power as much as they are by cost (or rather, money allocation), and due to the corruption of the scientific funding system in most rich countries, waste is extremely high, severely limiting the money actually going into science. Eg to make radical progress in physics, we need more sensitive telescopes, larger particle accelerators, more efficient space missions, more actual iteration on fusion reactors etc. All these things cost a lot of money, yet the majority of it gets eaten by systems designed to minimize the money actually going towards the research work. This isn't even really getting into how the researchers are paid. I agree that AI researchers are concerningly blasé about what they're actually aiming to do. AGI is poorly defined, and the only negative externalities they seem to pay lip service to is culture war stuff. There's near zero consideration for the real negative externalities they're perpetrating upon the world. I think partly this also manifests in how they interact with fields where even older AI developments would be useful. At the physics lab I work at, I've found that many scientists have a negative impression of AI 'researchers', where they don't care to understand the problem they're trying to solve, preferring to just treat everything - even things where the solution space can be constrained by scientific understanding - as a black box. It's almost the opposite of science in approach.

sebmellen 3y ago
It’s still surprising to me that not one decabillionaire has funded a good basic sciences institute. With a few hundred million dollars (paid primarily to salaries) and an extremely rigorous selection process, one could establish a highly prestigious research university and bring the brightest minds together in one place. Just go from there! If Leland Stanford could do it…
- rsanek 3y ago
  https://arcinstitute.org/ https://arcinstitute.org/
  - sebmellen 3y ago
    Arc is a cool concept, but these initiatives are ultimately handicapped by their reliance on the infrastructure of "grants". Grants suck. What matters is having smart people together in one place. If you pay smart people enough money, and offer them enough insulation and protection from academic politics, they will come to your institute. Grant-making to people at institutions which are already politically and socially captured does not fix the root issue.
- timeforcomputer 3y ago
  Not sure if relevant as it is not just one institute but the Simons Foundation (https://en.wikipedia.org/wiki/Simons_Foundation https://en.wikipedia.org/wiki/Simons_Foundation) seems along this line.
timeforcomputer 3y ago
I am studying physics now (with a CS and math background) and I feel obligated to get up to date on AI and to develop a good working philosophy of how it can be meaningfully used in scientific work. Not that all "apply machine learning to X" approaches aren't interesting, but I lack enough understanding to know whether these are popping up everywhere because people feel obligated to apply new methods. For example, the Fourier transform is deep and interesting and there are libraries and standards and ways to transform different objects over clusters, etc., but I wouldn't say good scientific research is about finding a place to apply a Fourier transform (maybe :)). I am new to this though.
- dotnet00 3y ago
  I'm nearing the end of my PhD in computer engineering, but because my uni is associated with a big physics research lab, my work ended up being pretty involved with physics. I've had to go through a similar learning experience regarding how to meaningfully apply AI (and computation in general) to perform meaningful science. I agree that simply applying a fourier transform isn't meaningful research, but there are fields like fourier optics, where you're effectively just doing that, with the goal of modeling wave optics. The kinds of issues I had in mind wrt poor uses of AI are things like inverse problem solving methods which train a model as a black box and neglect to involve physics informed feedback, so the results are of questionable value, or hyperparameter estimators which need an unrealistic amount of data or produce blatantly unrealistic estimates because they're naive and the AI "researcher" has no interest in actually understanding the physical logic to the parameters. My research hasn't been with AI, just with simulations of certain physical systems, and one thing that was constantly drilled into me by the physicists was to be really careful about how far I stretch computational techniques, because ultimately none of this digital stuff means much if it becomes detached from reality. It can even cause actual damage because sometimes expensive purchases will be made on the basis of simulations.
  - timeforcomputer 3y ago
    Thanks for the pointer to Fourier optics! I mentioned Fourier because it is something I still don't "get" even after studying and using these methods. I am still learning, but I have so far only applied inverse methods to systems where the physics is well-defined and the model already accepted as "real". For example, CT scanner algorithms which model the photon counters and attenuation through matter (although this can get complicated...), with the output being data in well-defined format (like the distribution of matter). I do see people using inverse solvers or AI to derive something which isn't an "image" but a model of the physics (does that make sense?). The extreme version of this which I have seen once before, is a paper using AI to extract algebraic forms of PDEs from videos of fluids. To what extent does the AI user have to understand/"pre-model" the physics, and to what extent does the AI generate understanding in some way? Apologies if this is misinformed as I am still outside the research community, and don't very actively read papers. Currently I am trying to learn background to understand electromagnetic properties of materials with an interest to apply this to computer graphics. I eventually want to learn what processes are currently done for extracting material properties from scans, etc., along the lines of the classic gonioreflectometers. In this space, from the papers I look at, it looks like AI is unavoidable and I am sure an extremely useful tool but I definitely want to be careful not to get lost in misunderstanding.

sebmellen 3y ago

It’s still surprising to me that not one decabillionaire has funded a good basic sciences institute. With a few hundred million dollars (paid primarily to salaries) and an extremely rigorous selection process, one could establish a highly prestigious research university and bring the brightest minds together in one place. Just go from there! If Leland Stanford could do it…

rsanek 3y ago
https://arcinstitute.org/ https://arcinstitute.org/
- sebmellen 3y ago
  Arc is a cool concept, but these initiatives are ultimately handicapped by their reliance on the infrastructure of "grants". Grants suck. What matters is having smart people together in one place. If you pay smart people enough money, and offer them enough insulation and protection from academic politics, they will come to your institute. Grant-making to people at institutions which are already politically and socially captured does not fix the root issue.
timeforcomputer 3y ago
Not sure if relevant as it is not just one institute but the Simons Foundation (https://en.wikipedia.org/wiki/Simons_Foundation https://en.wikipedia.org/wiki/Simons_Foundation) seems along this line.

timeforcomputer 3y ago

I am studying physics now (with a CS and math background) and I feel obligated to get up to date on AI and to develop a good working philosophy of how it can be meaningfully used in scientific work. Not that all "apply machine learning to X" approaches aren't interesting, but I lack enough understanding to know whether these are popping up everywhere because people feel obligated to apply new methods. For example, the Fourier transform is deep and interesting and there are libraries and standards and ways to transform different objects over clusters, etc., but I wouldn't say good scientific research is about finding a place to apply a Fourier transform (maybe :)). I am new to this though.

dotnet00 3y ago
I'm nearing the end of my PhD in computer engineering, but because my uni is associated with a big physics research lab, my work ended up being pretty involved with physics. I've had to go through a similar learning experience regarding how to meaningfully apply AI (and computation in general) to perform meaningful science. I agree that simply applying a fourier transform isn't meaningful research, but there are fields like fourier optics, where you're effectively just doing that, with the goal of modeling wave optics. The kinds of issues I had in mind wrt poor uses of AI are things like inverse problem solving methods which train a model as a black box and neglect to involve physics informed feedback, so the results are of questionable value, or hyperparameter estimators which need an unrealistic amount of data or produce blatantly unrealistic estimates because they're naive and the AI "researcher" has no interest in actually understanding the physical logic to the parameters. My research hasn't been with AI, just with simulations of certain physical systems, and one thing that was constantly drilled into me by the physicists was to be really careful about how far I stretch computational techniques, because ultimately none of this digital stuff means much if it becomes detached from reality. It can even cause actual damage because sometimes expensive purchases will be made on the basis of simulations.
- timeforcomputer 3y ago
  Thanks for the pointer to Fourier optics! I mentioned Fourier because it is something I still don't "get" even after studying and using these methods. I am still learning, but I have so far only applied inverse methods to systems where the physics is well-defined and the model already accepted as "real". For example, CT scanner algorithms which model the photon counters and attenuation through matter (although this can get complicated...), with the output being data in well-defined format (like the distribution of matter). I do see people using inverse solvers or AI to derive something which isn't an "image" but a model of the physics (does that make sense?). The extreme version of this which I have seen once before, is a paper using AI to extract algebraic forms of PDEs from videos of fluids. To what extent does the AI user have to understand/"pre-model" the physics, and to what extent does the AI generate understanding in some way? Apologies if this is misinformed as I am still outside the research community, and don't very actively read papers. Currently I am trying to learn background to understand electromagnetic properties of materials with an interest to apply this to computer graphics. I eventually want to learn what processes are currently done for extracting material properties from scans, etc., along the lines of the classic gonioreflectometers. In this space, from the papers I look at, it looks like AI is unavoidable and I am sure an extremely useful tool but I definitely want to be careful not to get lost in misunderstanding.