A compiler and a assembler are predictable enough. An LLM will never be reproducible. Period. LLM base d AI's are doomed from the start.

You hate to see that kind of ignorance at this level of play. Wouldn't such sentiments be better expressed elsewhere, on sites not populated by users who are actively doing what you say can't or shouldn't be done?

Think whatever you want. Unless you are Donald Knuth level educated, you can't be 100% sure.

I think it is more likely that the "amazing" things, as described, will become mundane, thus shifting the definition of what becomes interesting or noteworthy. Even then, I believe people who continue to write by hand will still be appreciated, given this is fundamentally a hobbyist space.

One good example is nocash, who continues to write all of his Windows-targetting emulators in hand-rolled assembly to this very day. For a long time, some of his emulators - most notably NO$GBA - were at the forefront of popularity among homebrew developers due to their unmatched debugging tools; even though one could make them much faster with an HLL, nobody really stepped up to the task. Even now, his work remains relevant thanks to the extensive documentation he maintained alongside the emulators he developed.

NO$GBA is written in assembly language?! I can't believe it.

> All no$gba program code, both emulation and user interface, is all hand-crafted and pure assembler code.

https://problemkaputt.de/gba-dev.htm

Proud to be a Luddite.

Less so in the hobby area, more so in the "they're taking our jobs" area.

Do you make your own shirts and pants, too?

Actually, one of the larger youtubers is currently facing massive backlash from his own community because all his projects are done by the slop machine :)

The great thing about retro computing is we want to see what people come up with. Not slop bots!

To be precise, it's people using AI.

I massively disagree. If someone uses AI to port a whole game engine to $RETRO_OS and that means I can port a great game to that OS by hand that I wouldn't have been able to do otherwise, everybody has won there.

I hope most people are like Obama's energy strategy, famously described as "All of the above."[1] -- the AI-powered efforts don't take away anyone's ability to do amazing "by hand" accomplishments. There's so much in the problem space that there's no chance of all the interesting projects being "solved with AI." I want everyone interested to be building things, with or without AI as they prefer.

[1] https://obamawhitehouse.archives.gov/sites/default/files/doc...