Raw LLM Responses

Sorry but the opposite is true. It is nice looking from the outside to say see it is all bad. AI is evil, but seeing how fast the coding tools are developing I hear from professionals all around me in every part of the IT space that they are experiencing the same as myself. AI coding tools are making huge leaps forward, where as little as 6 months ago this article might have been accurate today's tools are so much closer to what their human counterparts can do it is scary. No more code duplication, way better maintainability providing that the person prompting the tool knows how to tell it to produce good code. This in turn requires someone that knows good code so they can recognize when the AI is being silly, it does still get silly sometimes when incorrectly prompted. But with a half decent developer at the controls these tools are by now making the work faster and maintenance easier, producing tests for an entire exiting large codebase is a mind numbingly boring task that AI can do for developers. This in turn allows the developers with or without the help of AI tools to then make code changes safe in the knowledge that if all tests still pass afterwards everything will work as before. I am sorry to say but there is most certainly a realization in the IT world that AI is not (yet) going to just replace all developers or the vast majority of developers. But the speed with which the tools are improving and becoming more capable is shocking and the general consensus is that these tools will replace the vast majority of developers, with a human and the latest and greatest AI tool that work together to produce more code in less time. Code quality will not be better or worse just like with a carpenter and their tools a master craftsman will produce better results than a untrained beginner given the exact same tools. The thing is though the agents that are now being build are much more specialized and are able to work together to produce better results than they would on their own. We have seen this in the AI labs for a while already but now we are seeing this more and more out in the wild where a group of agents each specialized in one particular thing is able to work together to produce a much better result than a single general purpose AI agent can. This is the next step in this story. Why would a human look at the code a AI generated and evaluate it's maintainability? Why not have an agent do it and feed back the results of it's review to the coding agent. Why would a human perform the repetitive QA testing on yet another version of the same software why not have a specialized agent do that? And so on and on. The state that this report is talking about has already changed massively and is very rapidly changing all the time. In the next year it is likely that we will see the first large corporations launch new applications that are mostly AI written, and even AI architected. The question is not if developers are going to lose their jobs to AI agents the question is when. Though the main effect that less developers will have is that there will be new roles to essentially keep the same number of people now simply with a new set of tasks compared to the developers of old.