The funny thing about this is that it makes perfect sense if you know the history of coding languages/practices and you trace through how things have changed.
The earlier programming was done in languages that were as close as possible to machine language, and as such you never used memory or resources that you didn't need. But it took fucking forever because it was meant for machines to parse, not for humans.
Then as time goes on, each iteration of "New" programming languages moved farther away from "This is machine language" and towards "This is human readable". And it allowed us to do more ambitious things with software at the cost of performance. At first, the cost was small to negligible. But as we followed this trajectory, the bits of memory and processing power that we were let slide got bigger and bigger in favor of more powerful programming languages.
Then, Programming was pushed as the end-all be-all for a prosperous career, and so a lot of people who had no interest in the reality of computer programming began to fill all the jobs at these big companies and churn out less and less precise code
But now, we're giving people a tool that allows them to have the ease of access that powerful-but-inefficient languages have, while being able to be much more focused on things like Memory usage and CPU. AI is much more capable of writing machine level code, or using powerful languages in the most efficient way possible.
I think we're about to discover that our computers are FAR more powerful than we realized after AI starts optimizing all the software slop we've been served over the years.