For as much as I like what Mythos can do, I think Steve’s a little overconfident in its abilities to declare that we’re now in some post-vulnerability world where software like Firefox will simply be perfect because an AI scanned through the code. Or maybe I’m just biased because there’s code in Firefox that’s almost as old as I am, and wish they wouldn’t have abandoned their Servo project.
Still, based on 20 years of listening to this show I can’t help but think we’re all being a little too trusting of software that scans other software to declare it has no bugs anymore. It’s really supposed to be that simple?
When Microsoft has 2 Patch Tuesdays with nothing critical I’ll start to believe. Until then I remain skeptical. I say this as someone who’s going to vibecode himself out of a job within the next 2 years.
The old saying of you can’t prove a negative applies. A scanner saying it can’t find any bugs does not mean there are not any to be found. On the other hand, the machine has a bigger context than us humans, and theoretically it doesn’t get tired or distracted, so it really is a task better suited to machines than to humans.
I think it will change the game a lot and, over time, the number of serious errors will diminish, we will clear a percentage of errors each time it is checked, but you never get to zero. Plus you are putting in new code all the time, yes, checking the changes with AI will help avoid building in new errors, but I don’t think that it will catch 100% of errors, or new types of errors that it has never seen before…
We will see a huge surge in bugs initially and it will decline over time, but zero errors? I doubt it, most companies, unlike Steve, are under pressure to release new versions on a regular timescale and that means that corners often have to be cut and sorted out later..