Raw LLM Responses

@Pammie right now its no replacement for programmers. for instance if you tell it to write code that deletes NANs in a pandas dataframe, it will do it. even though in some cases it is better to interpolate over NANS. what Im saying is that its only as smart as the user. it cant read the dataframe and then tell you exactly the best steps. its the same with code, if you dont tell exactly how to program or use objects or inheritance or any of that, it will do dumb things. this means only advanced comp sci students who know algorithms and software best practices can use chatgpt to programe effectively. a lay person trying to make an app with chatgpt will fail. and yea, this LLM model stuff will improve. but theres a reason people have limited tokens of context. these things take massive amounts of computing power. the trained model takes less power than training, but its still some fraction. gpt5 or 6 will take an obscene amount of computing power. so much so it simply wont be feasible to have everyone using it with unlimited context tokens, not effectively. without some major breakthrough in chip technology or quantum computing (not even sure current training or software methods would work in quantum) the LLM model is fast approaching the physical limitations of modern computing. tl;dr seriously no need to panic. I would guess in about 15 or 20 years a programming job will be replaceable (by something better than scaled up LLMs) but youre totally safe for now, youve got a lifetime ahead of you to prepare and learn new skills. as tina says, keep learning and youll remain relevant in any job market. frankly nuclear war wiping out society should be a bigger concern to you (youll need to learn to chop logs lol) then any AI taking low level input jobs