The Copernican Revolution (discovery earth was not at the center of the solar system) initially had worse empirical calculations because they didn't know planets traveled in ellipses.
The moments after the revolution might be worse, but in the long term, we got better.
It could be if it actually lets us calibrate our credence of your original claim that most revolutions have resulted in a lot of death for little benefit. If the worst examples are much worse than the best examples, or vice versa, then we can plausibly conclude whether you are at least directionally correct.
You don’t take the single best and single worst examples of a thing that has occurred thousands of times to determine if the results are more positive or negative on average.
That depends entirely on the specifics. If the worst revolution killed 20 people and the best led to the Enlightenment, scientific progress, vaccines, etc., you absolutely can judge your claim's merits, even if there were 10,000 examples of those bad revolutions and only 1 example of a good revolution.
Do you think this is a productive exercise and we are going to A) answer this and/or B) glean anything from it? This back and forth feels does not feel productive, unless we want to playfully and in good faith game out how this could maybe be done.
Forest fires are immensely destructive, but they clear the way for new growth in their wake. The same has been said for recessions and the economy, and I think there's at least some comparison to be made for revolutions and societies.
Nah this revolution the billionaires who control the Ai and automated means of production will voluntarily give their money to the little guy instead of needing widespread unrest and riots beforehand like the other times
For the case of Propositional Logic, ChatGPT reflects the current epistemological crisis. When asking for help on a question, it could not properly apply the Law of the Excluded Middle [1].
I would love for this to turn out to be some internal constraint where the LLM can not ‘reason’ about LEM and will always go to an understanding based in constructive logic. However, I am more ready to accept that LLM aren’t actually ‘reasoning’ about anything and it’s an inherent flaw in how we talk about the algorithms as though they were actually thinking ‘minds’ instead of very fancy syntax completion machines.
The problem is that both constructive logic and "normal" logic are part of the training data. You might be able to say "using constructive logic, prove X". But even that depends on none of the non-constructive training data "leaking" into the part of the model that it uses for answering such a query. I don't think LLMs have hard partitions like that, so you may not get a purely constructive proof even if that's what you ask for. Worse, the non-constructive part may be not obvious.
Your comment is certainly correct and I agree that the various implementations of LLM probably can not actually partition attempts to find proofs into any given logical system.
My comment was more tongue in cheek than your response. I phrased it awkwardly obviously. I was humorously hoping that at some fundamental level, involving unintentionally taking advantage of a previously unknown foundational rule of computer science, that LLM’s as ‘thinking’ algorithms simply could not understand or utilize non-constructive logical means to formulate a proof.
As I said, I did not think this is actually what’s going on with GPT not being able to actually, or convincingly, ‘understand’ the law of the excluded middle. It was more of backhanded insult at LLMs, particularly, and those sales people who want to talk about them as thinking, reasoning, semi-conscious algorithmic ‘beings’.
As a Brazilian, I believe the problem is that we have a culture of "gratitude" towards the government. We quietly and silently thank God that we received whatever benefit and pray that they will keep giving us that. But a tiny bit of economical education, and an open eye to the frequent corruption scandals teaches one that there is more than enough money for a decent salary for academic workers.
I haven't used them in big codebases, but they were also able to help me understand the code they generated. Isn't this feasible (yet) on big codebases?
Yesterday I spent the entire day working on a lib to create repos in Github from inside Emacs.
It was the first time in 3y that I had touched it.
When googling, I saw potential candidates that were much better than my simple one.
But I kept going, for the pleasure of making my own thing.
I learned a lot, and felt very accomplished, even if, at the end, it was messy, and I'll have to go back and reorganize it.
It feels like making _my_ thing, even if it is drawing my copy of Monalisa.
Intelectual work like writing, researching, teaching, etc. despite being important, don't have intrinsic appeal such that people naturally and voluntarily put money on it. We are not built like that.
This is where institutions like universities, governments, etc. come in.
Of course you need to be good at selling in situations where you can get a good cut. I’m sure there are a lot of car salesmen that are just squeaking by.
But, yes, good salesmen for things like enterprise software can do quite well though you more exposed of the vagaries of the market than someone more removed from the front lines.
> This is where institutions like universities, governments, etc. come in.
Science was doing pretty well before it became institutionalized in the early 20th century. It's not without tradeoffs, but these aren't essential components.
