It’s a bold strategy, Cotton. Let’s see if it pays off for them. https://www.tomshardware.com/tech-industry/big-tech/cisco-gives-27-year-old-engineer-unlimited-ai-tokens-to-rewrite-98-percent-of-the-companys-legacy-code-in-24-months-president-plucks-former-startup-founder-telling-him-to-go-nuts
@Meyerweb paging @davidgerard did you see this? Amazing. Gonna go so well! "4" auth tokens everywhere!
@Meyerweb I know his cofounder, and I suspect the "something unstrategic" was something on the team that I left shortly before they were acquired. I can verify that Jeetu has successfully created tension in the teams, but it's not really about ideas either. He has regularly communicated that anyone not "working on AI" doesn't belong in the company, which meant every team was trying to independently rush a chat bot POC out into their product, without any quality standard.
@Meyerweb lol, even AI can't make this shit up
@Meyerweb Cisco and making bad decisions before a likely seismic economic bubble popping, name a more iconic duo.
@Meyerweb they won’t need to build a backdoor.
@Meyerweb I fully assumed this was from a satire site at first.
@Meyerweb
This can't possiblie go wrong
@Meyerweb To give Cisco a tiny amount of credit, converting legacy code is one of the things that AI does the best at. Where AI really goes off the rails is with vibecoding; somebody says, "Make me an X!" and the AI has to hallucinate an entire application into being.
For legacy code, it has the existing code that it can refer back to and verify that the new system behaves the same as the old, so it will never stray too far from what it's supposed to be doing, which is just the kind of thing that improves AI's chances of getting things right.
And then to judge them realistically, this is still going to go tits up because even if the AI was good 99% of the time, and the legacy code was 100% good, that's 1% of the code in critical networking infrastructure that's going to be vulnerable and nobody's going to understand the code base anymore.
The AI will not be good 99% of the time.
And I don't believe for a second that Cisco's legacy code is 100% good.
The intersection of "AI is being naughty" and "the code it's converting has a bug or vulnerability" is of particular interest. It's highly unlikely to fix bugs, and best-case scenario it reproduces them. But, using bugged code as a basis for AI-driven development can propagate even more bugs throughout the system, like building a wall and being "just a little bit" off at the bottom.
I'd say to get off Cisco while you can, but I'm honestly not sure where you'd go that's not slop-pilled.