LLMs predict the next token probabilistically.
If you donโt know the word or lack domain knowledge, itโs extremely hard for them to generate anything useful.
A lot of people think they can just dump everything on an LLM and solve it with a single click.
Thatโs not how it works.
You actually need to know how to do things yourself if you want to handle 100k concurrent users or solve real hard problems.
The people who blindly trust probabilistic LLMs and skip studying will be the first ones replaced.
Use LLMs as leverage. Study. Build real thinking skills and problem-solving ability.
Still donโt want to study?
Bro, I'm so goddamn fed up pretending this isn't a problem.
the internet spent two decades building tutorials, open-source software, blog posts & answers for free.
AI companies took all of that and turned it into billions of dollars in products.
and now these innovators and open-source contributors are being told they're disposable.
We made the library... someone profits off the entrance gates.