There's a cartoon going around where in the first frame, one character points to their screen and says to another: "AI turns this single bullet point list into a long email I can pretend I wrote".
And in the other frame, there are two different characters, one of them presumably the receiver of the email sent in the first frame, who says to their colleague: "AI makes a single bullet point out of this long email I can pretend I read".
There's a trend of people replying to posts/tweets/etc. with 'I asked ChatGPT and it said...'
It's the modern equivalent of LMGTFY. The OP could just as easily written the same prompt themselves. The difference is that LMGTFY was an expression of irritation, smugness and hazing. The ChatGPT reply is just garrulous laziness. I expect and hope we'll develop social rules that mean this type of reply will be seen as passe.
The last time I got one of those lazy ChatGPT responses I wanted to just ban the person on the spot if I had moderator privileges. Just pages of dreck that looked like detailed information but was totally useless and a waste of time. I don't have a problem if people use ChatGPT and find it helpful, but it's hugely disrespectful to just copy and paste its output to other people without even a cursory review of it first.
tl;dr: AI is looking to convey words. A good author is looking to efficiently convey information.
Because that's literally what it is. Its an algorithm that is continuously asking itself, 'what is the most likely word I should say next?'
Whereas an author that is intending to communicate a point, will start with an idea, write a passage to explain the idea, and then edit their passage to the minimum number of words that most precisely, accurately, and succinctly communicates that idea.
The most scary thing nowadays is that a lot of people fail Reverse Turing Test. They think that the thing there (the LLM) is thinking... They say "I told chat" etc. If high-ranking people will fail the test they are able to start using AI instead of thinking, or for example impose laws using AI...
Same here, but AI is orthogonal to that. Spam is spam - there's no difference between one written by silicon-based LLM bot, and one written by protein-based low paid human bot.
This is untenable. I could be AI. You could be AI. The whole idea of value is going to change when there is 99.99% noise from AI, and genuine human created content will be hard to distinguish if at all.
1: people will use ChatGPT to write their formal emails based on a casually written text
2: people will use ChatGPT to convert their emails from formal text to summaries\
3: this will get automated by email providers
4: eventually the automation will be removed and we'll just talk in plain language again
I do #1 and it's great. I'm not sure why folks in this thread are sneering at what is a working English transpiler. I don't get to choose having to write formally at work sometimes but I can control how I deal with it.
I'll buy the AI LLM that finds the prompt that was used based on the output, and replaces posts and emails by their prompt so I can actually understand the intention of the author and not spend my whole afternoon going through all bullet points with highly redundant information.
Not that /s, really. If you think about it, what a person writing a long-winded e-mail full of redundant text is doing, is the same work LLM is - they have a prompt in their mind, and they're generating text that "sounds nice" out of it.
AI or not, it would be better if they just sent their prompt instead.
The process is extremely lossy, but guessing a likely prompt should be doable if they used an LLM just to inflate word count, to reverse it, just remove redundancy and noise. Saves you time.
Then why did you even write more than two sentences in the first place, if nothing else matters? Why didn't you write a summary-line at the beginning, consisting of two sentences?
AI will not replace human thinking, even though many people seem to believe and put their brain on stand-by.
It feels like someone wants to transport water from A to B and transports it as steam, just because _we can _.
This is the really hilarious ‘engineer thinking’ vs ‘normie’ thinking difference which rears its head sometimes.
after all, what’s the point of a giving someone a bunch of cheap flashy gifts for Christmas (instead of say, socks) either?
As long as we can all pretend they were thoughtful and meaningful, and someone isn’t using AI when making it (or just picking random crap off the shelf, and they removed the price tags) or using AI when reading it (aka making a big show of opening it, and then throwing them in the trash immediately after the person leaves), then we all get along. It even looks like we’re doing a ton of work/spending a ton of money to make the other person happy.
Not that anyone does any of the things I’m describing, just being hypothetical, obviously.
I suspect it will be obvious enough shortly it will go the way of the ‘popcorn bucket’ fad or the like, but for now…
"Normies" actually prefer to get a paragraph long email rather then three pages saying the same thing. AI is NOT adding just a few socially expected niceties. It adds huge amount of fluff.
And what "normies" do with that is skimming it, ignoring majority of it and answering random part.
> And what "normies" do with that is skimming it, ignoring majority of it and answering random part.
Exactly that. For me, a lot of effort in structuring e-mails goes into making it look like text instead of bullet points, because some stupid social expectations, but then still making it bullet-pointed in nature, because if I don't, the typical normie recipient will do exactly what you said: ignore majority of it and answer random part.
(And then they'll somehow screw it up anyway, and I'll still have to chase them after that one critical question they conveniently forgot to address.)
> after all, what’s the point of a giving someone a bunch of cheap flashy gifts for Christmas (instead of say, socks) either?
Making them feel good and "seen", obviously. This is perfectly expressible in "engineer thinking" (I won't say "quantifiable", because there's this meme that engineers see things in binary, whereas the reality is, math is perfectly fine with fuzzy ideas and uncertainty - it's the normies that can't handle those).
Hell, there are some game-theoretic approaches to maximize social ROI on gifts, but I won't go into those, especially that they tend to flip the sign on the return if the recipient learns about them.
My email is disliked due to its brevity, turning the single clear and concise sentence of into a multi paragraph treatise might just lead to promotions, raises and bonuses which I can trickle down through the economy.
I think this underrates how many emails are literally just replies of "sounds good". Small snippet replies seem to be the vast majority of automatically suggested responses in gmail
A reply of "sounds good" means the initial email has been read and its contents agreed upon. Ho would AI improve upon this?
- sending "sounds good" even when the recipient hasn't, in fact, read the initial email => catastrophic alternative
- writing an elaborate email explaining in luxurious details why it in fact sounds good => not catastrophic, but costing time on the other side to read and understand, with zero added value
It's a paper trail for me. Companies, as we saw recently, can do whatever they want on company chat platforms. Emails are nearly impossible to fully delete if they ever have to escalate to a lawsuit, and can (YMMV based on policy) let you BCC important trails to your personal email.
If it's that important you can screenshot it. If you're BCCing every email you sent to your personal email that is (or should be) an IT policy violation.
It’s no different than using IMAP or POP3 to download your messages. This is the beauty and curse of email. It’s sometimes too transparent. I prefer it.
Yeah. It's not every email (I can probably count the number of times I did this on one hand). But if I feel like they're trying to bury some lead or simply want to CYA, I will try to at least download the email to the local machine (perfectly legimate) and BCC myself (Grey area) as an immitation of 3 backup strategies. I've never had to utilize thar BCC, fortunately.
But yes, phone screenshot is another strategy with much less grey area. I'm just becoming more and more paranoid of some potential defense trying to accuse my photo of being doctored, especially with more and more AI tools available.
Why is chat not a paper trail? Just yesterday I found a chat message that I had written in 2019 and I was surprised that I already back then knew things I did not know yesterday.
(We are use zulip for chat which is better than everything else I have used since irc. But the search is too limited for someone who knows regexes.)
Many reasons. First, chat doesn't exist. What exists is scores of incompatible chat apps.
I use WhatsApp but I consider WhatsApp messages throwaway because I keep losing them anyway. They are scattered across multiple phones with no way to merge them. Backups are platform specific. Exports don't contain any metadata and can't be imported.
"Chat" is a useless mess, not a paper trail.
For email, I have consistent backups with metadata across many email providers and email clients going back to 2008.
This seems risky, I'm not a lawyer but BCC company emails to personal account seems like a nice way to pave a highway for the company's legal team to request court ordered access to your personal affairs.
According to most work contracts / NDAs you wouldn't be allowed to keep private copies of work email.
If you are willing to violate that rule or the message affects your work contract which you are of course allowed to archive at least in zulip chat that's very simple (for a software person). They have a straightforward REST API. IIRC you can even choose between markdown source and HTML rendered output.
Because e-mail is naturally self-replicating and not bound to organizational boundaries, in ways chat isn't.
Chat messages tend to exist in one place only (vendors' servers), with maybe a transient local copy that gets wiped over time, or "for privacy reasons" (like Messenger switching to E2EE, effectively wiping cached history on any device that went through the transition). Chat message is an object, it's designed to exist in a single place, and everything else is a pointer to it, or a transient cache.
E-mails, in contrast, are always copied in full. You send an e-mail to me, you retain an independent copy, I get an independent copy, and a bunch of servers in between us keep an independent copy too, even if briefly. I forward your e-mail somewhere, more people and servers get their copies. I reply back to you, more independent copies, that also quote the previous messages, embedding even more copies that are even more independent. This makes it very similar to paper correspondence (particularly when photocopy machines are involved), i.e. impossible for a single party to unilaterally eradicate in practice.
And then chat vendors implement silly features like ability to retroactively unsend a message, force-deleting it from recipients' devices too (it may still exist in backups, but vendors refuse to let you access those, even with a GDPR request). In e-mail land, that's fundamentally not possible.
(Microsoft tried to bolt it onto their corporate e-mail software, but it only works in Outlook/Exchange land, and it's easy to disable (at least was, in OG Desktop Outlook - not the still broken New Outlook Desktop Web App). I discovered this when I once saw an e-mail I was reading suddenly disappear from my Outlook, which prompted me to find the right setting to disable honoring unsend requests.)
So, come discovery time, critical chat history may turn out impossible to find, and any deeper search will require forcing cooperation of the chat operator. E-mails, on the other hand, tend to turn up, because someone, somewhere, almost certainly has a copy.
Chat is a paper trail in finance at least. For regulatory purposes, bank personnel are not allowed to delete even their WhatsApp and other text messaging app info from their phones.
"serious business" and "serious stuff" still happens over email, and in the same way, even "more serious business stuff" happens over snail mail still.
Yeah, the reactions are just email messages with special headers, which as you say ends up spamming people who don't use Outlook. I think the hate was a mix of reaction to bad implementation and the concept in general.
I've adopted the inbox zero approach. If it's important it gets reclassified onto my task list with start and end dates, if it's useful info it gets filed, and everything else goes into trash.
At this point I am thinking my Thunderbird should probably just unify the Inbox view and the Task view, since it would be a more accurate representation of how I view email.
I thought just now, isn't inbox zero just a cosmetic difference?
For you: important things become tasks, useful things are filed, and everything else gets trashed.
For me: important things get opened and replied to. Useful things are starred (and opened). Everything else stays untouched.
And that pesky unread number is irrelevant because I mute all notifications. I'm not discounting your method, I am just now realizing the circle of it all.
There's a UX difference: when you look at your inbox from fresh you have to remember which ones you purposefully ignored because they were left unread in the inbox (this might be trivial for you if you're used to it).
I practice inbox zero also, the value for me is knowing that if it's in my inbox it's because it requires actioning, if it's not it's ignored (deleted or archived).
I also just generally like deleting things as much as possible, I don't like the cruft. If I have to search through old emails I don't have to filter by stars or anything like that, I like knowing that if it exists it's because it's important.
They specifically allow you a grammar/spell check and also change tone (formal/informal) and length. Length one I have never used but the grammar spell check is a godsend that I use almost always.
Yes, I'm aware. What AI/Proton provides isn't just a simple spellchecker though. It specifically recommends and alters wording to better suit the overall sentence structure. Essentially, it considers the context better than any built-in checker I've had in the past.
It's also really useful to for words that are spelt almost the same. Suit and Suite for example.
Also throughout my day, I'm constantly switching between 2 languages that have almost identically written words. Adress and Address. The normal spellchecks often don't mark it as an error because my computers and browsers naturally also have 2 installed keyboards and languages.
Maybe you aren't in a space where it would be useful, but not everyone who has to write an email is a great and concise writer.
I worked with groups of tradespeople who had poor literacy and they had to write emails and some of them were very poorly written. AI would have helped these people a great deal in providing information but also being able to understand what was coming back to them.
It almost can't be a good thing. LLMs are only useful when given all the relevant context. When you write an email, the context is mostly in your head.
It isn't, though; it's in all the meetings that happened beforehand and all the documents around them.
The biggest productivity boost I ever managed was using Whisper to convert meetings to text and then a big model to summarize what happened.
Then I can chat with the docs and meetings about who decided what, when, and why. It's a superpower that I could only implement because I'm in the C-suite and could tell everyone else to get bent if they didn't like it—and gave babysitters to the rest of the C-suite.
Having visibility and ownership for decisions is a huge deal when everyone has access to it.
This was about a year ago so it was Claude 3 Opus for summaries and interrogation. Since then pretty much anything over 70b is good enough.
And baby sit means hire something between a secretary developer that makes sure that important meetings had the record bot invited, gave it a once over and then went back to the 70% of their job that was actual development.
My experience with LLMs expanding on bullet points is that they often enough misrepresent my intentions as a writer. Often in infuriatingly subtle ways.
Same when summarizing, just less frequently.
As someone who cares about precision and clarity in my writing, I do not use LLMs in the context of communication.
Both frames of the cartoon represent a real perceived need: for the sender, the need to inflate the message to "look nice" because "people expect it", and then for the recipient, the need to summarize the nice-looking message to get the actual point they care about.
Hopefully the use of AI in email will make that cartoon (and the underlying message) widespread, and lead to people finally realizing what they failed to realize all these decades: just send the goddamn bullet point. We don't need AI in e-mails. We just need to stop wasting each other's time.
EDIT: and riffing off rpigab's comment downthread, https://news.ycombinator.com/item?id=42723756 -- I wish for the future people will feel comfortable, instead of sending AI-generated e-mails, to send the goddamn prompt instead. It carries all the information and much less noise.
I mean, using LLMs makes sense if you actually need to communicate in prose - for many, myself included, it's much easier to evaluate whether some text sounds right, than to write it that way in the first place, so LLMs are useful in evolving and refactoring your own writing (and learning how to write better from it, over time). But that is rarely the case in transactional or business communication - for that, just send the prompt.
Google seems to have an advantage here; as the client on both ends in many emails, they could just check if this ai expand/summary process is occurring and if so just send the bullet point (or if they want to be really clever just pass the bullet point through a thesaurus, so nobody will notice even if the sender happens to see what the recipient got).
what are people even worried about here? they're just trying things to see whether they're useful. don't expand your emails into long prose if it adds no value for you and they will focus on other things.
Really? Wow. And they think if they're pointing it out, it absolves them somehow? Like those companies that used to have Dilbert cartoons pinned on cubicle walls?
There's a cartoon going around where in the first frame, one character points to their screen and says to another: "AI turns this single bullet point list into a long email I can pretend I wrote".
And in the other frame, there are two different characters, one of them presumably the receiver of the email sent in the first frame, who says to their colleague: "AI makes a single bullet point out of this long email I can pretend I read".