If you told someone 2008 that the fun scrappy startup offering Gmail would be reading your emails for ads* (see reply to this!) until 2017 and only stopped when another tech company had a massive leak of personal information which helped sway a presidential election* people would call you absolutely mad.
Emails were sacred back then and it took Google and long and slow creep to face little backlash when they started reading our emails. I see no reason why listening devices will be any different.
If the information is available and valuable to Google (and me talking to my spouse about buying a new computer next week sure is!) they will slowly walk towards getting it. The steps are pretty clear:
1. Have listening device which only listens to "OK Google" mass adopted
2. Increase capabilities for EASILY defensible safety reasons, listening to alarms, then screams, then personal threats, etc
3. Face backlash over 2. and easily defend it because who wouldn't want to save children from burning buildings etc.
4. Public begins to paint privacy minded folks as crazies for wanting to burn children alive.
5. Repeat 2 through 4 a few times.
6. Ease into that ad listening because everyone either thinks you do it already or they just don't care because they're exhausted hearing people raise alarms for 2.
I'm not saying this is a nefarious management plan to ease into surveillance either. Google's MO is put money and time into making a feature useful then eventually monetize it. Making "OK Google" listening devices useful in ways the public finds acceptable is where they're at now. Eventually they will have paved the way for it to be profitable and palatable for the public when it's always listening to us.
* Google changed this policy right around Cambridge Anlytica time