deleted by creator

I am an actual (senior) software engineer, with a background in ML to boot.

I would start to worry if we were anywhere close to even dreaming of how AGI might actually work, but we’re not. It’s purely in the realm of science fiction. Until you meet the bar of AGI, there’s absolutely no risk of software engineering jobs being replaced.

Go or Chess are games with a fixed and simple ruleset and are very suited to what computers are really good at. Software engineering is the art of making the ambiguous and ill-defined into something entirely unambiguous and precisely defined, and that is something we are so far from achieving in computers it’s not even funny. ML is ultimately just applied statistics. It’s not magic, and it’s far from anything we would consider “intelligence”.

Also, LOL at “how much code development is straight up redundant”. If you think development amounts to just writing a bunch of boilerplate as though we were some kind of assembly line putting together the same thing over and over again, you’re sorely mistaken.

It was impossible for computers to beat chess and go masters when the computers were trying to play like humans -trying to model high level understanding of strategy and abstract values. The computers started winning when they got fast enough to brute force games - to calculate all of the possible outcomes from all of the possible moves, and to choose the best one.

This is basically the same difference between LLMs and ‘true’ general AI. The LLMs are brute forcing the next line of a screenplay, with no way to incorporate abstract concepts like truth or logic. If you confuse an LLM for an AI, then you’re going to be disappointed in its performance. If you accept that an LLM is a way to average past communications, and accept that a lot of its training set were fiction, then it’s an amazing tool for generating consensus text (given that the consensus includes fantasies and lies). It’s not going to write new code, but it will give you an approximation of all the existing examples of some algorithm. An approximation that may introduce errors, like copy-pasting sequential lines from every stackexchange answer.

Computer graphics, computer game opponents, they’re still doing the same things they were doing decades ago, and the improvements are just doing it all faster. General AI needs to do something different than LLMs and most other ML algorithms.

Learning frameworks has never been hard, and frankly does not make up the majority of a developer’s job. Maybe you do it while onboarding. Big whoop. Any good developer can do that fairly easily, and LLMs are entirely superfluous. Worse yet, since they are so commonly confidently incorrect, you have to constantly check if it’s even correct. I’d prefer to just read the documentation, thanks.

A mature engineering organization is not pumping out greenfield projects in new languages/frameworks all the time. Greenfield is usually pretty rare, and when you do get a greenfield project, it’s supposed to be done using established tools that everyone already knows.A tiny fraction of a developer’s job is actually writing code. Most of it is the soft skills necessary to navigate ambiguous requirements and drive a project to completion. And when we do actually program, it’s much more reading code than it is writing code, generally to gain enough understanding of the system in order to make a minor change.

LLMs are highly overrated. And even if it does manage to produce something useful, there’s much more to a codebase itself. There’s the socialization of knowledge around it and the thought process that went into it, none of which you gain when using an LLM. It’s adequate for producing boilerplate no one reads anyway, but that’s such a small fraction of what we even do (and hopefully, you can abstract away that boilerplate so you’re not writing it over and over again anyway).

Sure, just like businesses have always wanted “no-code” solutions to their problems to cut out the need for software engineers. We all know how that turned out. There was no threat then, and there’s no threat now.