Extremely sarcastic take but - shouldn't the 10x productivity boost of LLMs like their own in-house Gemini make maintaining software easier?
Nothing in terms of feature release / quality / maintenance I am seeing out Big Software would lead me to believe there's been any productivity increase.
They can't even add "respect the user's preferred language" to the AI assist prompt in Google Search - I've reported it countless times. Turns out, looking up local places/names/etc, does not mean I speak that language. They can't seem to add light/dark support to Gmail despite prefers-color-scheme CSS media feature being present in browsers for at least 6 years). I have zero remaining faith in Google's ability to do anything.
No. If your engineers can produce more ROI working on another product, it does not matter that now they are 10x more efficient. They will also be 10x in the other, more worth it product
I assume the support range is determined by firmware updates. I know ChromeOS struggled with supporting devices whose components no longer received support.
Google's internal tooling for this is actually way behind what's available to the general public (they just got access to Claude though!)
There would still need to be money to make from doing so and a will to do it.
My company has a shitty old defunct group management portal that has been twice superseded and is finally getting shut down. But we have LLMs, why not maintain it in perpetuity? Because it's shit, nobody likes it, there are other options, it would still cost someone's time just to follow all the processes we have to do releases and keep things running, security testing, yada yada.
Adding more years of support to some shitty old thing is certainly going to cost them more than it makes them.
Gemini is good only at benchmarks. In actual usage it sometimes approaches chatjimmy in quality
Apparently, you aren't meant to use data, measure anything, or look at the actual end result of how software is really being released. Line Go Up