When most of your time is spent ensuring your material needs are met, then whatever time remains is precious. Telling someone hustling two different "gig economy" jobs sixteen hours per day to give a shit about the long term privacy implications of the stupid smart speaker he got his kid for Christmas is asking… a lot.
This is why change through popular political action peaked in the US in the late 1960s/early 1970s. People, on average, had more leisure time on their hands than ever before. More leisure time ⇒ more time to think, organize, unify, and give a shit beyond the immediate here & now.
It a business decision to use an LLM.
It's a business decision not to beef up the Alexa hardware to support an LLM.
These decisions are weighted against reputational damage by Amazon with the outcome discussed here.
I doubt they have ever ran machine learning model on those speakers. Mostly just some simple one like "hey alexa"
But I get your point that e.g. whisper is also very good and can run fast locally on edge devices. We also have recently phi4-multimodal LLM that is 5.6B params and is good enough to do inference locally on beefy smartphones. Problem is probably those smart speakers have really low CPU/GPU. Also LLM with audio input probably can be better re WER and take your mood/emotion into account when doing inference.
But my point is not just about the ease of running whisper, it's that the device is already doing transcription. Even if it's not optimal, it works.
If I would like to guess why amazon is moving to cloud it's:
1) They support only 8 languages right now - cloud LLM or even whisper can support like 50 languages pretty well. I was always dissapointed that couldn't buy Google Mini or Alexa or Apple Home for my mum because none of them speak Polish.
2) They want to provide good support or those less beefy smartspeakers that don't have much power and those still sell well.
3) They wanna move people to this new Alexa subscription that they recently announced or make people more subscribe to Prime.
4) Gather more voice samples so they can train as good multilingual TTS as elevenlabs.
how convenient.