It’s absolutely not, because AI will just make up stuff, even if the email is just gibberish.
I get assigned jira tasks where the manager turns his 8 word notes into a whole paragraph, except ai does not kow the context, problem or the product he is talking about and just generates some buzzword filler text that is 80% wrong.
If you don’t trust it you can still read the emails. If you do, you don’t have to read the emails. If want to confirm what it does say a few times until you can decide if you trust it or not, you can do that.
Then what’s the point? If I deliver some code that gets right maybe 4/10 then my boss would tell me to get fucked and never deployed. Yet when companies slap an “AI” sticker on the same bad results it its somehow worth billions and the next hot thing.
Bit more context that is needed:
Not for normal Gmail users, and also not for every email - only larger multi-reply ones.
I dunno, seems like an ok feature to me.
Doesn’t change the fact about lack of consent.
Lack of consent for what?
Please read the article again.
Google “read” your emails and feed them into their AI regardless, always have. They don’t need your “consent” to add features or make UI changes.
I’m assuming you actually mean it should’ve been off by default?
It’s absolutely not, because AI will just make up stuff, even if the email is just gibberish. I get assigned jira tasks where the manager turns his 8 word notes into a whole paragraph, except ai does not kow the context, problem or the product he is talking about and just generates some buzzword filler text that is 80% wrong.
If only you could still see the actual emails……
Oh wait, you can!
So it’s just useless noise and shouldn’t even exists just to waste energy? You need to pick a lane.
How did you get that from what I said?
If you don’t trust it you can still read the emails. If you do, you don’t have to read the emails. If want to confirm what it does say a few times until you can decide if you trust it or not, you can do that.
Then what’s the point? If I deliver some code that gets right maybe 4/10 then my boss would tell me to get fucked and never deployed. Yet when companies slap an “AI” sticker on the same bad results it its somehow worth billions and the next hot thing.
I think a summary of an email is going to be right more like 9.5/10 times.