We are at the very early stage of what these tools can do. There is currently a lot of hype around AI and its potential future impact. Expectations range from “all developers will be out of a job in 5 years” to “we’ve heard this kind of stuff before, it won’t matter”. Reality will likely fall somewhere in between.
For code generation, LLM tools are good at generating boilerplate code, but huge communities exist to make libraries and improve tooling, largely to avoid this type of code. Asking ChatGPT to write a sort function is interesting, but nobody writes sort functions. Time will tell how much of the gap between standard tooling and libraries can be filled with generative LLM models. However, AI is a big thing, including more than just code generation. When it comes to automation of UI testing, scanning code for potential bugs, and even smart API and backend construction, we don’t know what solutions these technologies will lead to. However, I feel very safe in predicting that these tools will be assistive in nature, and they won’t bring about the “end of code” and certainly not the “end of the coder”. If anything, demand for good practitioners will likely increase.
There are many non-tech industries that have a lot of room for improvement from a technology perspective, and more efficient development simply means more opportunities to work on problems that were previously cost prohibitive.
