Ask HN: Learning Code Is Dead? What do you think? Should we spend time reading Python, Rust, etc. books or focus on CS concepts?Are you reading any coding books after LLMs? |
Ask HN: Learning Code Is Dead? What do you think? Should we spend time reading Python, Rust, etc. books or focus on CS concepts?Are you reading any coding books after LLMs? |
Long answer: No. Really.
Why?:
Any person who has honestly studied from CS to deep learning knows that someone has to maintain that software and AI struggles in highly critical systems. Unless one wants bugs like this in production created by LLMs [0]
[0] https://sketch.dev/blog/our-first-outage-from-llm-written-co...
Dead? No. At least not in the near-term.
Changing? Maybe. Probably.
Importance of knowing CS principles / theory? Absolutely not dead.
I would have some reservations about telling a current high-school graduate or college freshman "Hey, plan on going into software development as a career" because of concerns about how much things are going to change in the longer-term. Basically, we don't really know the "rate of change" with regards to AI, and we can't predict all of the ramifications of continued AI improvement.
But those reservations aren't quite enough to say that coding is "dead" and especially not in the "here and now". People have done neat stuff with "vibe coding" and what-not, but when it comes to really advancing the state of the art and doing novel and interesting stuff, somebody still have to actually understand the fundamentals and know how to do this stuff from first principles.
Learn Code to debug AI code to help make it better. In the 1990s, Visual BASIC had Wizards to generate the skeleton of code, and the developer filled in the meat of the program.