Gmail’s AI-powered email summaries can dig through your inbox for you

UserIDAlreadyInUse

Ars Praefectus
4,017
Subscriptor
would be nice to know how likely it is to make up random portions of the email summary.
"My mistake! Your mother did not actually refer to you as a 'wanker' and your sister did not confess her love for your husband! Let me resummarize your emails! In fact, what really is intended was that your mother wants to remind you that Grandma's birthday is on Sunday and she wants a walker, and your sister's husband is the one in love with you. Sorry for any confusion. On your behalf I've scanned your Google Drive and sent images labelled 'Nudes, private' to your sister's husband. And your mother."
 
Upvote
150 (152 / -2)

nicholas.lecompte

Smack-Fu Master, in training
33
would be nice to know how likely it is to make up random portions of the email summary.
This is by far my biggest annoyance with how AI products are marketed and presented. Clearly the confabulation rate is greater than 0%, because an LLM that reliable would curb-stomp GPT-4o in most benchmarks and Google would be singing its praises. Surely Google has some sort of internal metric based on testing. So why not tell the class? Instead of specific numbers, we instead get condescending crap: "of course AI is not 100% reliable, every sophisticated user knows this. Apparently it will take some time for your feeble little pre-AI brain to use these tools properly."
 
Upvote
56 (56 / 0)

theOGpetergregory

Ars Scholae Palatinae
801
Subscriptor++
the user types, "Reply saying I want to volunteer for the parent's group event," hits "enter," and then the chatbot instantly, without confirmation, sends an email.

That seems less efficient than just typing, "I want to volunteer for the parent's group event" directly in an email but I'm not an AI expert.
 
Upvote
115 (115 / 0)

theOGpetergregory

Ars Scholae Palatinae
801
Subscriptor++
"Summaries" has to be among the most useless, if not the most useless thing these glorified chatbots claim to do.
But what if--and hear me out--the summaries might not only be useless, they might actually be wrong?

Would you be somewhat interested, very interested or very interested?

🙄
 
Upvote
83 (85 / -2)

rcduke

Ars Scholae Palatinae
1,873
Subscriptor++
This can prove to be a very useful feature depending on the applicable data privacy policies. But since this is coming from Google, I seriously doubt this can be used without agreeing first to let Google using personal email data for training and advertising purposes.
I would assume using Gmail in any fashion already auto agrees us to the T&C that lets Google read our inboxes. This is just an attempt to summarize the emails. We'll have to see if it's wrong or right.
 
Upvote
8 (9 / -1)
The trust isn’t there yet. So AI does this thing and I end up spending more time validating its answer than if I had done it myself. And if I do find mistakes, it makes me even less inclined to use it and even less inclined to trust it. By the time it gets good (if it does), everyone will hate it so much that they won’t want to use it unless forced to. Which might be their endgame.
 
Upvote
39 (39 / 0)

marsilies

Ars Legatus Legionis
22,514
Subscriptor++
This is by far my biggest annoyance with how AI products are marketed and presented. Clearly the confabulation rate is greater than 0%, because an LLM that reliable would curb-stomp GPT-4o in most benchmarks and Google would be singing its praises. Surely Google has some sort of internal metric based on testing. So why not tell the class?
We use Google Workspace at work, and have used it since it was Google Apps for Business. An account rep recently tried to sell me on Gemini vai email, specifically mentioning the ability to generate "accurate summaries of emails." I asked how accurate it actually is, where's the studies, the numbers. etc. They replied that all they could say is that it's "highly accurate."

They then mentioned a Satisfaction Guarantee where we could cancel "at any time." I asked how that would work with the annual license, and they replied back that we'd only be able to cancel at the annual renewal date. I then asked how that jived with the "satisfaction guarantee" they mentioned, then half-jokingly asked if Gemini had drafted that part of the email. They replied that actually, yes, Gemini had written the completely false Satisfaction Guarantee text in their email, and sheepishly admitted that it's "not 100% accurate."

They stopped trying to sell me on the product after that. I wonder why.
 
Last edited:
Upvote
133 (133 / 0)

ZippyPeanut

Ars Tribunus Angusticlavius
16,499
The trust isn’t there yet. So AI does this thing and I end up spending more time validating its answer than if I had done it myself. And if I do find mistakes, it makes me even less inclined to use it and even less inclined to trust it. By the time it gets good (if it does), everyone will hate it so much that they won’t want to use it unless forced to. Which might be their endgame.
Yeah. I hate smartphone. but I'm forced to use one. I hate subscription-based software, but I'm forced to use it. I hate smart cars, but in 20 years I'll be forced to drive one. Yeah, you'll be using AI, in some fashion or another, whether you want to or not, in the near future.

And, yes, it will get good, scary fucking good.
 
Upvote
-14 (10 / -24)

Pecisk

Ars Scholae Palatinae
652
After 30 years of email summaries, better UI / UX, etc. it is very simple conclusion.

No amount of summaries gonna help. There are just too many emails. There is too much stuff we want to do. We can't help it it is how our brain works. Counterintuitively, but best way to handle it is disengage.
This is why none of Microsoft and Google innovations does not stick or move a needle. Are they nice? Sometimes yes. Microsoft finally has solid separate task application, which syncs with Exchange. But it breaks any mold? No. Summary or not, stuff is just TOO MUCH.
 
Upvote
9 (13 / -4)

marsilies

Ars Legatus Legionis
22,514
Subscriptor++
Yeah. I hate smartphone. but I'm forced to use one. I hate subscription-based software, but I'm forced to use it. I hate smart cars, but in 20 years I'll be forced to drive one. Yeah, you'll be using AI, in some fashion or another, whether you want to or not, in the near future.

And, yes, it will get good, scary fucking good.
Change your argument to Bitcoin/cryptocurrencies, NFTs, or VR, and you can see how flawed it can be to view any emerging technology as "inevitable." Just because major corporations are tripping over themselves to spend billions in this new field doesn't mean it's anything that's actually useful or will be adopted by the vast majority of people.
 
Upvote
45 (48 / -3)

Fatesrider

Ars Legatus Legionis
21,799
Subscriptor
I have a hard time imagining something I would want less.
My thought was, "Oh, good! Another reason to continue to use an e-mail client and never sign in to Google to read my e-mails!"

They reinforce that decision with every move they've made with Gmail for the last 20 years and more.

And, IMHO, if you're SO busy you can't be bothered to read and sort your e-mails yourself, well, I'd say some life priorities need to be rethought.

When you are forced to lie and say "I skimmed it" is one of those "you fucked up!" moments in all circumstances. You're trusting a machine that has demonstrably made up its own shit in often spectacular ways with your communications life. That's not just a bad idea. It could result in some pretty nasty repercussions.

The more Google fucks with the soup, the less I want to taste it.
 
Upvote
17 (18 / -1)