That presents an interesting privacy conundrum. It would undeniably be an incredible capability for you to have, but pretty much all of your day-to-day interactions would be sent off to Google or Microsoft for processing.
And as a techie I am not too bothered by Google and MS using my data to sell me stuff.
What bothers me much more about is the dark architectural decisions that misguide/divides people or get them addicted to their streams and devices screwing their lives up.
It's not your privacy that's the issue (Well, it is, as a user of the device) - it's also the privacy of the persons speaking with you. Single-party consent for digital recording of audio is a controversial topic.
But it's also the privacy of everyone close enough to the microphone that single processing can extract what they are saying. That's 0 party consent that I'm not ok with for always on devices.
Now this piqued my interest, so I took a look at Wikipedia[1]. Interestingly enough, England seems to allow recording phone calls without notifying the other party as long as it is only for personal use and not made available to a third party, but US Federal law does not seem to have such a stipulation, thus presumably allowing people in one-party consent states to post recordings of phone calls on social media without the other party's consent (unless the state forbids that. It's a bit hard to find concrete information on this), which is pretty troublesome I think. However that is only for telephone recordings and not for private conversations, so maybe the laws are a bit more sensible in that case.
Governments have a lot of work to do and that is part of 'how things should be' for people to be equal. I don't think 'people deserve things' but enabling people to be the best they can is government responsibility.
I'm not even sure which outcome I think I prefer, let alone which outcome is right.
It seems unfair for a hearing person to deny a deaf person the right to "hear" if he has a technological solution available, but it also seems unfair for the deaf person to be streaming everything the hearing person says to the cloud for analysis.
Maybe the deaf person can stream whatever he wants to the cloud, but you can refuse to talk in his presence if you don't want your speech to be streamed to the cloud?
The societal reaction to Google Glass was overwhelmingly negative, because it turns out that people don't want to be knowingly recorded all the time. Throw in some people that are easy to "other" and you have a pretty ugly situation on your hands.
Cochlear implants and hearing aids are available, and do work for people; throwing networking into the mix sounds really unnecessary, in the same way that a smart fridge talking to the internet does.
I completely agree that a speech-to-text tool for the deaf shouldn't be doing any networking, but in a (hypothetical or not) world where the only viable speech-to-text equipment is doing networking, what do we do?
Legislation focused on people with hearing impairments is exclusive and discriminatory. Legislation needs to instead focus on data use and storage.
Hearing loss is socially isolating and has tremendous impact on quality of life and opportunity. Only 20% of Canadians with profound hearing loss have full-time employment; 40% are unemployed and another 40% are underemployed.
Throwing "networking" into the mix, if it brings essentially real-world captions into the world for people with hearing loss, isn't on par with a smart fridge talking to the internet. It's a profoundly life-changing breakthrough that will improve the quality of life for 5-25% of the population that have some degree of hearing loss (2).
1. http://cad.ca/issues-positions/employment-and-employability 2. https://www.chs.ca/facts-and-figures