You ever read something and suddenly a whole classification scheme lights up in your head?
A thread on X from “stringking42069” showed me a combination of opinions I hadn’t seen before. stringking42069 is a pro-string theory commentator with a macho gym bro memer gimmick. He’s openly contemptuous of many physicists who describe themselves as string theorists, arguing that only a smaller number really deserve the name.
To be clear, none of that is the new combination. Long-time readers of this blog will remember a frequent commenter with a very similar attitude, if much less tendency to use the word “bro”.
The new thing, from my perspective, is how he thinks about AI. As he explains in that thread, he sees AI as great at certain kinds of physics calculations, ones where the methods and goals are mostly known and the challenge is working out the math. He doesn’t expect it to be able to contribute real creativity or judgement, the messy decision-making that physicists use to decide what is worth building in the first place.
Others with that perspective tend to argue that this will be a boon for scientists, who AI will free up to do creative work, multiplying their output. The difference is, stringking42069 thinks a lot of scientists are not doing creative work in the first place, including most of the people making extensive use of AI. So if anything he’s happy to see them go, and only pissed that they’re sucking up resources and attention on the way out, and discouraging students who could be joining the parts of the field that do real creative work.
It made me realize that there are two axes to thinking about AI in physics.
On the one hand, there’s where you think AI capabilities are. Is AI going to lead to “a nation of geniuses in a data center”, an AI-powered super-(cyber-)Ed Witten for everything and everyone? Is AI great at routine work and coding, but will never be able to do anything really creative or novel? Or is AI total hype, almost always a waste of time?
On the other hand, there’s another axis: misanthropy about science. For some of the people arguing about AI online, most scientists are good people trying their best to do worthwhile things. For others, most scientists are complacent and cliquish, wasting time and money on ideas that are going nowhere and forcing the real geniuses out of the field.
Put those together, and you get the table below:
| Thinks academia is mostly fine | Misanthrope | |
| AI geniuses are coming | The practice of science will change. We’ll play at science like chess, and have fun trying to read and understand amazing AI insights. | Soon all scientists will be out of a job when the public notices AI can do it all better. Then the real breakthroughs will come. |
| AI can do routine work | AI frees scientists to focus on what we do best: creativity. We should think carefully about how to train junior scientists now, though. | AI is comparable to bad scientists who only do derivative work. If they leave, we real paradigm-changers could inherit the field. |
| AI is complete hype | Most scientists don’t use AI. AI is worrying because it misleads students and the public, who should listen to real scientists. | Scientists are shilling for AI companies, as you should expect for people who waste the public’s money on reputation games. |
This classification is missing a lot, of course. One important question is not just what AI can do in principle, but what it can do cost-effectively, and whether anyone is actually willing to pay for it. A point where I agree with stringking42069 is that companies get a lot of good PR out of building AI physicists right now, and that PR benefit won’t be relevant forever. I’m also leaving out the more general questions of AI’s effect on society, for example people who think AI geniuses will lead to the end of the world as we know it.
But I suspect if you look at this table, you can already start matching the scientists you see on social media. I’ve seen examples of all of these in the wild (though the bottom-left is somewhat rare, as far as I can tell). Where do you fall?
4 gravitons 
I am firmly in the ‘Sorcerers Apprentice’ block.o much hype
So much hype from sources that are projecting along a single axis
LikeLike
It’s too much hype now but that’s short-term; longer term, AI geniuses are inevitable; we just just don’t know how long a term.
Put a high tax on tokens and put its product under safety review.
LikeLike
From a psychologist perspective, the binary outcomes glosses over alternative perspectives that the outcome will most likely be middling; a kind of best of the worst, worst of the best of each position.
From what little I’ve read, AI (algorithmic information, not artificial intelligence), algorithms predict an output based on probabilities, but have no understanding of what it means, which is what intelligence implies; understanding.
Of course, the caveat applies that I’m not a computer scientist, just a retired psychologist who understand cognitive bias in people who are scientists.
LikeLike
I think the key question is what difference understanding makes in the outcomes. I can see a nuanced take between the middle row and the bottom row, with “AI will be able to do certain specialized types of tasks, but not others”. Much harder to prognosticate about, but probably closer to true!
LikeLiked by 1 person
I’d fall sharply in the middle row but I think both columns. I try to automate routine work at a reasonable price in my day job using LLMs and don’t really see a way around having human beings read and understand everything we do regardless.
LikeLike
I note that you tag this post “Machine learning”. I’m not sure that the large language model generative AI that writes papers with a disturbing share of hallucinated citations but reads like a professionally written physics paper is all that similar to a machine learning program.
Machine learning is usually used to describe software that is typically given a big data set and told in vague terms to either come up with functional relationships that describe the data (e.g., come up with a function with no more than four degrees of freedom the best describes inferred dark matter distributions in a data set), or to bin the data in a way that flows from the bottom up when the task of categorizing data points is non-obvious (e.g., classifying hundreds of thousands of galaxy images into different categories of galaxies).
Below “machine learning”, in turn, is “dumb” software that simple implements specific rules provided entirely by a programmer, in an extreme case, with little more sophistication than a Microsoft Excel spreadsheet, and in the less extreme case, something like the Mathematica software package that solve very challenging equations according to somewhat more sophisticated instructions and sets of tools to try.
The problem, to a minor extent with machines learning, and to a major extent with generative AI large language models, is that they aren’t transparent. You put input into a black box and it gives you output, but you have only a dim idea (if any) of the “reasoning” or operational decision making tests that it applied to get there.
Is it really science if you can’t prove how you got your results and may not even be able to understand the answer?
That’s an issue that Douglas Adams was already grappling with in his Hitchhiker’s Guide to the Galaxy series decades ago.
LikeLike
An LLM is a functional relationship that describes the data (the data being token orderings in texts), it’s just one with an absurdly large number of parameters. If anything I don’t think anyone would describe simply fitting a few-parameter function as ML, that’s just statistics as ordinarily practiced. Now if you’re trying to come up with an arbitrary four-parameter function that’s different from trying to fix the parameters in a particular four-parameter function. But as a result, you have a lot more parameters, and the more complexity you’re trying to get out the more you need.
As to the question of whether the output of a black box should count as science, we are of course all black boxes to one another. We mitigate that by consistently explaining our reasoning. LLMs don’t seem to be good enough at that yet, but there are a wide range of possible answers as to why, some that imply they will never be good enough and some that imply they will be any day now.
LikeLike