manfromchina1

It wasnt like that several months ago. It and then-GPT suddenly swapped personalities. GPT has loosened up considerably and now makes more mistakes while Claude has become what it is while also smarter.

Does anyone have to understand AI output when it comes to math? Then that someone has to be trained in math. Yeah, for all intents and purposes that person is a mathematician. Maybe not the one who gets money for that in which case they are not a professional mathematician. Maybe these paid guys will disappear. However if AI keeps churning out new stuff nobody understands humanity will have to create a new category of employment for those who do to make sense of all that. Mathematicians, then, are back in the game again.

One possibility is that AI keeps churning out new math and funnels it into other branches of science and engineering without any human input. At which point math becomes a black box. Maybe in that case human mathematicians might become obsolete. But then other branches will start becoming incomprehensible as well.

> Sometimes I’ll have an argument that goes A->B->C, and the model will suggest I reorder as B->A->C. If I try that and feed it into a new instance of the same model, it’ll sometimes say “that’s great, but I suggest ordering it as A->B->C”, and so on forever.

Sonnet 5 acts like that currently. GPT used to be like that, but it looks like they shaved off that unnecessary behavior somehow. But GPT is now also kind of lazy and pretends like it has done its research when it hasnt.

Why not? We still have to learn how to write. AI is much better at it than the average guy anyway. It can do any style in most any language. No human on Earth is capable of that.

You want to learn programming in C. Pick up Linux skills. Read CS:APP by Bryant / O'Hallaron and OSTEP by the Arpaci-Dusseau. Should be a good basis for kernel hacking.

I personally don't get this attitude. For one thing, nobody on earth can give you a 500-word summary of quantum physics in a way that lets you internalize even a tiny bit of it. That's not how you learn anything.

Here's how I do it: I open Baby Rudin (3rd ed.), second chapter, and read the main text - absolute brutality. I unpack almost every sentence with Claude/GPT until I finally get what's going on. No ELI5 nonsense, just examples and counterexamples galore while absorbing the techniques and the way of thinking in analysis/topology. How do I know I've learned the material? By solving every single problem in that chapter. Here's the thing, though: the problems in Rudin can be brutal and decoupled from what's in the text, so if you can handle them, you've definitely mastered the material. No 500-word summary of analysis here.

What kind of bidness?

It does, but it also sounds like one of the older GPTs - much higher and sturdier guardrails now. While do I also get the sense that GPT has loosened up a little?

Depends on what kind of profs.Those who teach low level systems stuff absolutely need you to be able to program in C or something similar. On the theoretical side you can kinda sorta get away with very strong math skills. You open CLRS - it just gives you a pseudocode, you do efficiency/correctness analysis on it and forward you go. Then open an Automata book and realize it's all math all the way down. Not a lot of math either.

Kernel programming still resists AI so C and assembly arent going anywhere.

I rarely go to the front page.

27 is a Tao prime. Terence Tao suggested 27 was a prime number on The Colbert Report in 2014. He was likely very nervous.

Kernel Hacking. Human judgement here is still required and irreplaceable. For fast employment - Non Destructive Testing.

Keep studying physics, though. Go up to modern Solid State physics for a possibility of entering industries that revolve around chip production.

How far along are you? You could finish your phd, then get a two year diploma in mechatronics, electrical eng, instrumentation. All near future proof and apparently pay well enough. You could even cert up in NDT, IRATA, ROV and that sort of thing. With your phd you'd get your foot in the door pretty easily.

Even ChatGPT does that.

This extends to alien bio as well. I took a lengthy write-up from reddit about supposed alien physiology and asked Claude to summarize it for me. It was flagged. It went thru when I added a text about how it was speculative and not at all illegal.

Free ChatGPT is like a fast car with a barely responsive steering wheel. Guardrails on that thing are insane. Even for math. It wont let you think. It will try to fix mistakes you havent even made yet based on intent that was ascribed to you for no reason. It veers off in some crazy directions thinking that's what you meant and trying to address even a little bit of that creates almost a combinatorial explosion of even more wrong things. Is why I stick to Claude. The latter is chill and only addresses what you had typed. Isn't verbose and actually asks you what you getting at with your post. That said, ChatGPT is more technical and can easily solve math problems that stump Claude.

Despite my handle I am not Chinese lest I am accused of harboring any bias. Anyway, I decided to look up what is going on. Apparently, things are nuanced. On the one hand China lags behind when it comes to semiconductors, large commercial aircraft, some pharmaceutical innovation. On the other hand,

> When evaluating the top 10 percent of high-quality scientific publications, ASPI finds that China surpasses the United States across all 8 critical technology domains. The gap is particularly pronounced in the energy and environment domain, where China accounts for 46 percent of top-tier publications compared to just 10 percent for the United States. Despite U.S. leadership in AI, China produces more top publications, contributing 30 percent versus 18 percent for the United States[0]

Basically, China dominates in batteries, solar, quantum communications, robotics deployment, high-speed rail, nuclear construction, autonomous vehicle deployment, manufacturing process innovation, patent volume in most categories

[0] https://itif.org/publications/2025/09/23/how-china-is-outper...

In the meantime Grok seems to have come a long way. Its free tier allows you to upload 350 page pdfs without complaint and actually crunch data for you. It's now much better at it than either Claude or GPT. It has also become a much more capable mathematician. I am blown away by the new Grok.

I drove from New Orleans to NYC through the Blue Ridge Parkway once. I thought it was extraordinarily beautiful. Also drove through the north of Arizona from LA back to NOLA. I was surprised how northern the terrain looked. All alpine forest and no cacti or anything even though when you take I-10 from Florida to California it's all cacti, canyons and mountains. Very beautiful in its own right. The scenery in New Mexico could easily shatter minds from Maine down to Florida on the eastern seaboard. It's eerie even during the day when you're driving through that Martian terrain.

What about the converse? Are the stars of conservatories of music particularly good at math and such? Interestingly, most music here is geared toward millennials and older gens.

Families who insist you study 12 hours a day for higher SAT scores are also likely to believe in 'correct' sort of music and enforce it until that's all you like.

> More relevantly for us, Iran is 3.5 times larger than Iraq and roughly twice the population.

Worth noting that at the time of invasion of Iraq they had about 25 million people per gemeni. They now have about 46 mil people per wikipedia. All else equal, we are comparing 25 mil to 93 mil and not half of 93 mil to 93 mil.

Reminds me of "Total Rickall"(Rick and Morty: Season 2, Episode 4) which itself is probably a retelling of a retelling of a retelling of a retelling.

I personally did the following:

1. Ploughed thru every problem in "Schaum's Outline of Programming With C++". It's an old book, but the code in it is not particularly different from modern C++

2. Picked up "C++ Crash Course: A Fast-Paced Introduction" to fill in the gaps, understand what wasnt explained in (1) etc

Sounds like how America was 20 years ago.

A US visa is typically for entering only. It doesnt say anything about the duration of stay. The latter is determined by CBP at the port of entrance and printed on your I-94. This is never emphasized anywhere.

The newest one on the CIA channel on YouTube is showing 1920 views right now. It was posted 3 hours ago.

He still won.

They would likely pick the cards according to what your text looks like at the moment. You should at least ask them to use a random function. And even then they might keep picking according to the mood/content of the text.

That link has the following: https://x.com/nettermike/status/2009843044028428714

> Security Guard: Without a doubt. I'm sending a warning to anyone who thinks they can fight the United States. They have no idea what they're capable of. After what I saw, I never want to be on the other side of that again. They're not to be messed with.

Sounds like a strategic bluff.

> I haven't been anywhere in the world that has LA beat.

I feel the same way about NYC and NOLA. Both have world class cuisine and the sheer diversity of NYC food scene can't be found anywhere in the US. And for whatever reason NYC style pizza is not as ubiquitous outside the NYC-NJ area.You can definitely find such pizza in Boston, but it cannot be found in every little corner in LA for example.