LLM Dilemma, How to be ahead of curve in changing tech landscape?
So Within 3 months, I unsubscribed both GPT Premium and Claude Pro. While I found ChatGPT useful, I heard Claude is very good in writing code So I unsubscribed ChatGPT to use claude. Claude was good for brainstorming and overall system design but it doesn’t remember previous chats and even after giving prompts it maintain the same response.
Also found out that I am not learning anything by using LLMs and just becoming good in giving prompts where I will be easily replaceable.
Is it good to be in getting the work done without having expertise in long term because that’s what I do at my work, I get my work done?
Now since as a software engineer, I face this problem and then people say you have to take advantage of AI/LLMs to get ahead of the curve.
Now I really want to know how to be ahead of the curve so LLMs do not replace me. Any thoughts?
Finally: Which LLMs are useful?
I believe whatever code any LLMs generates ,you do need to understand it line by line. Even if it works flawlessly you don't want to have a piece of code in your codebase that you have no idea about (at least on the modules you are working on in case it's a big codebase).
Now the only way to be ahead of the curve is to avoid being an expert beginner. If you end up just doing prompt engineering without understanding the underlying concepts it's not going to end well. For example: It might give you a working implementation of a http server in your preferred language but it's your job to understand what all goes behind in making a http request: what protocols are used,how the packets are transferred, proxies etc.
Chat gpt 4 is good I will try out cursor.ai have heard good things about this.