It's sad to see that the sane opinion is so heavily downvoted.
LSP as a protocol is fine, but the actual technical implementation of JSON RPC is braindead. Only web devs that don't know anything about native code could devise such an abomination. What happened to plugins and dll's?
I love C++ for the power it gives me, but boy do I hate reading C++ code. I know most of these things are for historical reasons and/or done because of parser compatibilities etc. but it's still a pain.
A lot of it is about making metaprogramming a lot easier to write and to read.
No more enable_if kludges since if constexpr (and concepts for specific stuff); and using concepts allows to better communicate intent (e.g.: template<typename I> can become template<std::integral I> if you want the template to be used only with integer types, and so on)
Thankfully, you can still write C++ just fine without the "modern" stuff and have not only readable code, but also sane compile times. The notion, explicitly mentioned in the article, that all this insane verbosity also adds 5 seconds to your build for a single executor invocation is just crazy to me (it is far longer than my entire build for most projects).
It's not crazy, it's just what happens if you write mostly C with some conveniences where they actually make sense instead of "modern C++". I generally write very performance sensitive code, so it's naturally fairly low on abstraction, but usually most of my projects take between one and two seconds to build (that's a complete rebuild with a unity build, I don't do incremental builds). Those that involve CUDA take a bit longer because nvcc is very slow, but I generally build kernels separately (and in parallel) with the rest of the code and just link them together at the end.
Sure, C++ is heavy for compilation, there's simply more by the compiler to do, but code repository building under 5 seconds is at the very low end of tail so making the point about someone bearing with the 5 seconds longer build time is sort of moot.
I wrote a lot of plain C and a lot of C++ (cumulatively probably close to MLoC) and I can't remember any C code that would compile in such a short time unless it was a library code or some trivial example.
Two intelligence assets talking to each other. Both have quite similar backgrounds with dubious credentials. A history of lying and obfuscation. I wouldn't trust anything Fridman or Durov say.
> Math Academy is an AI-powered, fully-automated online math-learning platform. Math Academy meets each student where they are via an adaptive diagnostic assessment and introduces and reinforces concepts based on each student’s individual strengths and weaknesses.
What is edutech and why is it unsuitable?
I don't want a computer in the loop when I learn math, plain and simple. My preferred style of learning is instructor led with a mix of Socratic method and hand holding. But bar that, reading texts and using a pen and paper.
As far as I can tell, most of its value comes from having a reasonably thorough dependency tree of math topics and corresponding exercises (which can be solved with pen and paper) and describing it as "AI" is how you get investors to fund a math textbook.
The “ai” is an expert system yes to calibrate to your ability to answer questions it throws at you. The questions are all human written. I had your initial scepticism as well, I can reassure you that the ai is not an LLM.
Also the guy Justin skycak who built it has put a lot of thought into its pedagogy
My experience with MathAcademy is very positive. So is my experience using ChatGPT 5 as a math teacher in learning mode. I'm as fed up with AI slop as most people, but for me this is a domain where it excels.
I'm pretty sure the author hasn't actually followed his own advice. Also, the writing style was atrocious. I wouldn't trust anything said. Seems like the type of post he wrote for himself (nothing wrong with that), but I don't feel that asking a 20-25 year old to have a 5 year plan which is basically 1/4 of their conscious life is something useful or achievable for most.
I think as with everything related to learning if you're conscientious and studious this will be a major boost (no idea, but I plan on trying it out tonight on some math I've been studying). And likewise if you just use it to do your homework without putting in the effort you won't see any benefit or actively degrade.
Until that same AI starts shilling ads and certain viewpoints peddled by their owners in the output... This will happen 100% (ads, the other bit has already happened). The economics of all of these models doesn't work as is. There will be a major squeeze down the line.
Some of us have dipped our toes in local LLMs. To be sure, the ones I can run on my hardware always pale when compared to the online ones. But perhaps in time the ones you can run locally will be good enough.
Or perhaps an Apple or Kagi will host an LLM with no built-in monetization skewing its answers.
LSP as a protocol is fine, but the actual technical implementation of JSON RPC is braindead. Only web devs that don't know anything about native code could devise such an abomination. What happened to plugins and dll's?