LLMs are not 100% correct 100% of the time. LLMs are subjective. Code should work 100% of the time, be readable, and objective.
We also already have "easier ways of writing software" - website builders, open source libraries, StackOverflow answers, etc.
Software builds on itself. It's already faster to copy and paste someone's GitHub repo of their snake game than to ask an LLM to build a snake game for you. Software problems will continue to get more challenging as we create more complex software with unsolved answers.
If anything, software engineers will be more and more valuable in the future (just as the past few decades have shown how software engineers have become increasingly more valuable). Those that code with LLMs won't be able to retain their jobs solving the harder problems of tomorrow.
We also already have "easier ways of writing software" - website builders, open source libraries, StackOverflow answers, etc.
Software builds on itself. It's already faster to copy and paste someone's GitHub repo of their snake game than to ask an LLM to build a snake game for you. Software problems will continue to get more challenging as we create more complex software with unsolved answers.
If anything, software engineers will be more and more valuable in the future (just as the past few decades have shown how software engineers have become increasingly more valuable). Those that code with LLMs won't be able to retain their jobs solving the harder problems of tomorrow.