Linus Torvalds once said that LLMs could bring a tenfold increase in programmer productivity. But speaking at the Open Source Summit India 2026, Reported that this number “wasn’t scientific.” “This number came from me, obviously.”
Today, he continued, “we’re at the point where hopefully it creates more productivity than it takes away,” but “we’ve certainly seen more garbage being produced by LLMs than useful code by the beginning of this year… it can actually be a huge waste of resources when it takes a lot of effort from people to figure out that, that machine-generated report wasn’t real.” Even now, he said, “most of the good stuff requires more than just the LLM,” because “we had to push hard… if you find a bug with an LLM, it’s not enough to just ask the LLM to make a bug report and then throw it to us (for review). We want to see a proposed fix.”
Ο Torvalds described many AI-generated patches as “stupid junk patches… they might fix the immediate problem, but the bug kind of remains and just waits in the hallway to hit somewhere else.” For his own projects, he uses LLMs: “I use them as a way to create prototypes… quite often the code is not usable in that form, but it’s a great way to try something out,” while insisting that for kernel-level fixes, “LLMs, in my experience, haven’t gotten to that level yet.”
Torvalds acknowledged that some problems identified by AI were “surprisingly, I mean, interesting in a painful way,” especially security problems.
Finally he said:
“I think we’re much better at LLMs finding bugs, even when they’re annoying, and they’re things we probably should have found two decades ago.”
Although the press releases will range from very select to rare, I said I'd pass...because sometimes the editors hide.
