Aprender a programar con documentación o con IA
Posted by Foreign-Ant-4392@reddit | learnprogramming | View on Reddit | 1 comments
Últimamente presento un gran debate con la IA, estoy en 4to semestre de ingenieria de sistemas y las ideas de que la IA nos va a reemplazar es algo que me atormenta cada día (no es broma). Estoy tan obsesionado que he buscado toda la información posible sobre las consecuencias de uso de IA, y veo diferentes estudios que explican lo malo que puede llegar a ser su uso excesivo.
Por esto mismo, no quiero ser un inutil con conocimientos vacios, quiero ser un gran programador, realmente amo este arte y aunque no puedo negar que lo quiero hacer por dinero, la satisfaccion que siento al poder escribir código por mi mismo es inexplicable. He notado que al intentar estudiar con IA, no sigo un proceso estructurado sino que GPT me arroja fragmentos de conocimiento sueltos sin más. Por esto mismo, estoy pensando en desapegarme completamente del uso de IA y volver a métodos tradicionales como aprender un lenguaje de programación por medio de documentación e investigación propia. ¿Qué opinan?
amejin@reddit
Having lived through both sides of this - if you want to be "good" then read the docs, or at the very least learn to use them.
Programming and problem solving has never been about memorizing how something is done in a system, but knowing where to look for reference when checking if a process meets specifications.
LLMs can write code for you. LLMs can summarize specifications for you. In many cases, it can bridge the gap and write the code that meetd the specifications.
It cannot, to this day, write maintainable, secure, optimal code in one shot, nor can it know how your particular use case will apply the thing being built. An engine for a car is functionally the same as an engine for a truck, but there are differences and tolerances that one can handle better than another.
You being the expert means you can help guide an LLM to refactor, or often times you can just refactor faster yourself.
The industry is being sold snake oil right now, and a sucker is willing to see their stock prices go up so someone else can figure it out later. Across the board, anecdotally, I am hearing engineers complain about reduced staff, increased output, and a requirement to maintain consistency and stability.
This will end, eventually. The brain drain that occurs right now will be felt in a few years, and the cycle will repeat.