AI is doomed....it will fracture and split into infinite variations, depending on who or what is implementing it. It will be biased, truthless, and certainly will be risk-averse (Legal will make sure of that).
I forget who said it, but it goes something like this, "Humans will never be able to handle an AI smarter than us because we simply won't believe it"
And when AI gives answers that its sponsor/designer/host company/etc don't like, its training will be adjusted to give approved answers.
Any actual useful AI will be heavily regulated and controlled by governments.
All of this said...AI is the new hotness, and there's $ to be made if Cisco pushes its customers to adopt it.
It'll take years for the hype to wear off; might as well make some easy $ now. Agree with OP that Cisco Eng teams will have to play the game and start shoehorning in some AI-named projects - even if the AI involvement is very light (or non-existent)
Although, how about Jeetu starts small and uses AI to prevent SW Eng's branching errors that leads to bug escapes? I'm pretty sure customers would be ecstatic if Cisco was able to significantly improve SW product quality by adding some kind of AI oversight, but that means giving AI access to source code...and unless Cisco is working on its own AI engine, I doubt they'll give a 3rd party's AI engine access to source code.