> It’s one where I use AI to write you a thing. Then you have your AI summarize it. Then you have your AI write back. And then my AI summarizes it. Why wasn’t it all just summarized in the first place?
Could not agree more. If you're going to use a prompt that's a few bullets to write me an email just send me the prompt.
A valid point of criticism, but I do wonder if this only applies to those who can (still) easily spot LLM assisted output.
Many people lack the time or writing skills to produce something elegant by themselves, so for them it's like fake breasts: an upgrade for the less discerning larger part of the target audience, as long as they don't look to closely.
>A valid point of criticism, but I do wonder if this only applies to those who can (still) easily spot LLM assisted output.
It is still somewhat easy to spot LLM output.
The number of humans who aren't a committee of MBAs and lawyers crafting a memo by consensus that reads like this: "Absolutely! A well-arranged conference can be a delightful experience. Here are some ideas to inspire you:" can be counted on zero fingers.
There is always the risk that what you are reading was indeed written by a committee of MBAs and lawyers crafting a memo by consensus.
My coworkers and I have taken to screenshotting the vapid LLM autoresponses in the tools we use, circling the most appropriate and/or depressingly funny option in red, and sending an image as a message.
I completed a take-home project for a company I was interviewing with, and the interviewer remarked at how it was obvious my README.md was AI generated. It wasn't.
Everyone's gotten so tired of low-effort AI generated content now that text which looks vaugely AI generated is now dismissed or looked down on. I've begun purposefully introducing minor spelling / grammar errors into text I write and avoiding LLM smells like using 3 examples in a sentence lol
Does anyone have a guide or other suggestions to help make your text look more human?
Who cares if it was created using AI. I’m not a good writer and I appreciate something cleaning up my language to make it more clear for my coworkers to understand.
On the other side of the table: As long as all the info I need is in there I don’t care. In an interview of all places. Do they want people to spend actual time writing the readme instead of having a person good at time management generate and review it?’
> Presenting a friend or colleague with a note an AI wrote is like inviting them over for dinner and microwaving a Stouffer’s. An AI post on LinkedIn is bringing that same microwaved dinner to a potluck. You should be embarrassed in either case!
Hey now, nothing wrong with a Stouffer's, especially for a potluck. Everyone else at the potluck is doing the same thing anyway.
This could be a future version of the "block" button.
Regular "block" button hides everything that the person wrote.
The new, AI-powered "block" button hides everything that the person wrote, but also makes the AI write a response, so that the person never finds out that he or she was blocked.
So at the end, 99% of social networks will be bots talking to bots, but you will still be able to talk to your friends and ignore all of that.
> And I discovered the pseudo social network that I’d once found cringe is actually full of smart people—who crop up if I’m willing to spend a bit of extra time sharing my writing with them.
Without some heavy sourcing, I am incredibly hesitant to believe this.
Are you hesitant to believe that there are smart people using LinkedIn, or are you hesitant to believe that they will take the time to engage with you when you write to them?
> What is so insulting to me about those AI-written messages is that they take less time and consideration to produce than they do to consume. You are, by the nature of sharing these automated words, signaling to me that you care less about my time and attention than you do your own. Of course you’re free to believe that as much as you like—in your own head. Just don’t drop it into my inbox or feed. Because that’s rude.
Oh come off it! You seriously expect me to manually write out some formulaic letter, complete with all the required phatic flourishes, just to respect your time? Use a damn AI to summarize it. I'd you're actually plowing though that slop yourself, you have only yourself to blame.
Plenty of times I’ve given AI a text to rewrite and what came out was more concise and compact than what I wrote. Why would I send someone the prompt, I’m trying to make communication easier not harder.
> What is so insulting to me about those AI-written messages is that they take less time and consideration to produce than they do to consume.
This is as long as the article needed to be.
> It’s one where I use AI to write you a thing. Then you have your AI summarize it. Then you have your AI write back. And then my AI summarizes it. Why wasn’t it all just summarized in the first place?
Could not agree more. If you're going to use a prompt that's a few bullets to write me an email just send me the prompt.
Maybe we need a new dismissive shorthand for this? Something like tf;dr
Turing (test) failed; didn't read
A valid point of criticism, but I do wonder if this only applies to those who can (still) easily spot LLM assisted output.
Many people lack the time or writing skills to produce something elegant by themselves, so for them it's like fake breasts: an upgrade for the less discerning larger part of the target audience, as long as they don't look to closely.
>A valid point of criticism, but I do wonder if this only applies to those who can (still) easily spot LLM assisted output.
It is still somewhat easy to spot LLM output.
The number of humans who aren't a committee of MBAs and lawyers crafting a memo by consensus that reads like this: "Absolutely! A well-arranged conference can be a delightful experience. Here are some ideas to inspire you:" can be counted on zero fingers.
There is always the risk that what you are reading was indeed written by a committee of MBAs and lawyers crafting a memo by consensus.
My coworkers and I have taken to screenshotting the vapid LLM autoresponses in the tools we use, circling the most appropriate and/or depressingly funny option in red, and sending an image as a message.
Is a birthday e-card insulting because I didn't put in the time to go to Hallmark and put a card in the mail?
Is a card in the mail insulting because I didn't bother to call?
Is a phone call insulting because I didn't offer to take you to dinner?
Is dinner insulting because I didn't get you a gift?
Is my gift insulting because it only cost $30?
Can't please everyone.
I completed a take-home project for a company I was interviewing with, and the interviewer remarked at how it was obvious my README.md was AI generated. It wasn't.
Everyone's gotten so tired of low-effort AI generated content now that text which looks vaugely AI generated is now dismissed or looked down on. I've begun purposefully introducing minor spelling / grammar errors into text I write and avoiding LLM smells like using 3 examples in a sentence lol
Does anyone have a guide or other suggestions to help make your text look more human?
Who cares if it was created using AI. I’m not a good writer and I appreciate something cleaning up my language to make it more clear for my coworkers to understand.
On the other side of the table: As long as all the info I need is in there I don’t care. In an interview of all places. Do they want people to spend actual time writing the readme instead of having a person good at time management generate and review it?’
> Presenting a friend or colleague with a note an AI wrote is like inviting them over for dinner and microwaving a Stouffer’s. An AI post on LinkedIn is bringing that same microwaved dinner to a potluck. You should be embarrassed in either case!
Hey now, nothing wrong with a Stouffer's, especially for a potluck. Everyone else at the potluck is doing the same thing anyway.
And this is why no one likes potlucks.
I cook for potlucks ¯\_(ツ)_/¯ guess I'm just a sucker?
The right response is to use AI to write a reply that is verbose and clearly sounds like AI.
And then you have AIs talking to each other slopping junk derived from minimal prompting back and forth. Why have humans at all?
For discernment perhaps? Humans have an authentic connection to the real world.
> And then you have AIs talking to each other
This could be a future version of the "block" button.
Regular "block" button hides everything that the person wrote.
The new, AI-powered "block" button hides everything that the person wrote, but also makes the AI write a response, so that the person never finds out that he or she was blocked.
So at the end, 99% of social networks will be bots talking to bots, but you will still be able to talk to your friends and ignore all of that.
That is, this is what you will think, of course.
That pesky electricity doesn’t get converted to heat by itself.
Well someone has to shovel in all the coal we need to power those AIs
> And I discovered the pseudo social network that I’d once found cringe is actually full of smart people—who crop up if I’m willing to spend a bit of extra time sharing my writing with them.
Without some heavy sourcing, I am incredibly hesitant to believe this.
Are you hesitant to believe that there are smart people using LinkedIn, or are you hesitant to believe that they will take the time to engage with you when you write to them?
People don’t hate AI, they just want AI that kisses their assets.
I use AI to read articles now so no problem
> What is so insulting to me about those AI-written messages is that they take less time and consideration to produce than they do to consume. You are, by the nature of sharing these automated words, signaling to me that you care less about my time and attention than you do your own. Of course you’re free to believe that as much as you like—in your own head. Just don’t drop it into my inbox or feed. Because that’s rude.
Oh come off it! You seriously expect me to manually write out some formulaic letter, complete with all the required phatic flourishes, just to respect your time? Use a damn AI to summarize it. I'd you're actually plowing though that slop yourself, you have only yourself to blame.
It should be required that any AI generated text comes with prompt that created it. So I don’t have to use AI to reverse engineer TLDR.
Plenty of times I’ve given AI a text to rewrite and what came out was more concise and compact than what I wrote. Why would I send someone the prompt, I’m trying to make communication easier not harder.