It's been literally years since someone saying Alexa on a TV show or elsewhere have triggered my devices. Occasionally I've seen them switch to "listen", but they've gotten good at not reacting.
I'm assuming part of it is that it's gotten very good at recognising who is speaking. E.g. my son and I tested the limits of it a while back by changing voices (it still correctly identified my son when he tried to change his voice), mixing and matching (I'd say "Alexa" and he would say "who am I?" and vice versa; Alexa would recognise whomever said "Alexa" irrespective of who spoke what followed). So it'll usually know if whomever is saying "Alexa" is a known member of the household, which would seem like a good indicator to increase the threshold for how clearly the phrase is spoken before activating.
I'm pretty sure this is exactly how Google does it (only matching voices on the keyword); my Google home will ignore anyone else saying "ok Google", however if someone's yelling at me from another room or talking on the phone in the hallway, it's impossible to get anything useful out of it; it'll activate on the keyword and start processing whatever the first sounds it hears (dinner's ready, ok, ok, ok; I'm leaving now; you've got a phone call; etc). Before I have a chance to tell it to pause my music so I can hear the person yelling, it responds to them, not me. Before I can ask for the weather, it responds to someone walking by on the phone that happens to say something just outside my door at just the wrong time (this happens often when you live with people working from home)
I'm guessing there's something about how they're doing recognition of who is speaking that would make it hard to scale the cloud based speech recognition. Though I also notice that Alexa can answer more questions with the network down than it used to. Early on it was pretty much only the wake words, and then it'd given an error no matter what. Now it "listens" and will answer some requests to some degree without a network connection (e.g. if you ask to set an alarm it'll tell you the network is down and it can't set new alarms, but that it will still alert you, so it'd seem the full smarts of parsing an alarm request goes to the cloud, but it recognises at least enough to know you've asked about an alarm)
I'm assuming part of it is that it's gotten very good at recognising who is speaking. E.g. my son and I tested the limits of it a while back by changing voices (it still correctly identified my son when he tried to change his voice), mixing and matching (I'd say "Alexa" and he would say "who am I?" and vice versa; Alexa would recognise whomever said "Alexa" irrespective of who spoke what followed). So it'll usually know if whomever is saying "Alexa" is a known member of the household, which would seem like a good indicator to increase the threshold for how clearly the phrase is spoken before activating.