Software has been too complex for humans all along
Was software just always too complex for humans?

After AI-assisted code cracked the Riemann hypothesis with 17 million lines no human fully understands, Anil Madhavapeddy asks whether software engineering was ever really under control. His answer: better models alone won't fix it. He proposes deriving all software from formal specifications and deliberately engineering diversity across implementations, models, and harnesses so one bug can't take down everything at once.
I don't think better harnesses and models alone will let software engineering catch up, as they just produce mountains of slop code.