You did it 3 years ago? Your brain get fried in such a short time?
If you take a very broad definition of “using AI” as ever having used a Google search where an AI summary came up, or having used a Linux or Microsoft kernel, IDE or software component version where AI was used for coding or checking/verifying the code in any way, then yes, it’s almost impossible, as you will have to revert your tooling completely to a 2022 or earlier toolchain.
However, your decision to use an LLM (AI) to generate new code is fully your decision as a programmer to make, always will be and I don’t think that’s anywhere close to impossible.
Its impossible to hand roll code at the same rate as an LLM. Which means that if you don’t use them, you better have a good explanation for your manager as to why you’re clearing far fewer features/tickets relative to your colleagues.
You can try the “But my quality is higher/defect rate is lower”, but that gap is shrinking, and it doesnt seem like game studios even care about that anyway.
We all still have the skills to write code, its just that the people who pay us don’t actually care about human written code, they care about getting the game/software written as fast as possible.
You are right… though I will point out that is the speed of a delivering product which is what AI is good at and what managers want, speed of writing lines of code was never a good metric. Other than that programs have always been judged by their quality, function and performance, which perhaps AI can still improve and get better than the average programmer. Most tech companies have in the last decade prioritizing shipping fast over shipping quality, so that’s why many companies see AI as something that fit right in with their production philosophy.
Nonetheless, the point is whether coding without AI is impossible as Molyneux said depends on your definition, Pandora’s box is opened and it’s in almost every tool, but creating new code or documentation with AI or not is still your own choice.
It’s VERY possible if you’re not a hack.
Lol Peter Molyneux.
Still sad that someone has been making grand, unverifiable claims about the emergent behaviours of AI that utterly fail to manifest for the last few years and it was not Peter Molyneux.
rabble rabble rabble rabble!
Using VS Code, it’s impossible to tell what the recommendation or auto complete is from. Is it AI? Is it context of the current code? Is there a difference anymore?
Regardless, I write my own code and I use AI to teach me HOW to make functional code, since I’m poop at coding. I’m using as a tool for learning and self improvement.
