just apropos of this, one of the wildest things that i encounter in my daily life is how deeply ai tools are ingrained in so many people's workflows now. it's like in a movie when you realize that the other person is infected too. "so I asked chatgpt" NO
that bitch brainstorms its ass off, writes unit tests clean and can sometimes spot convoluted errors faster than i could read the code but yeah it becomes a horrible crutch a lot
Someone's gotta find a way to make it usable without rotting your ability to code
unfortunately that comes down to your individual will to live and be good at things so probably we are fucked
So odd that the will to live is maybe one of the weaker drives these days
this would require preventing the machine from generating code, idk how feasible that would be given that it seems they cant reliably stop it from doing anything
My wife's therapist suggested she talk to ChatGPT about something and I felt like reality jolted under my feet. Like it or not this is a Thing we have to deal with as a society now, and by god I wish we had a functioning legislature to help with it because we are doing a TERRIBLE job on our own
Work, shmirk, I’m seeing it just for life stuff man. Ppl going to it for medical advice. And you know, it might not be wrong but why take the risk?????
It's decent for getting me started on some task that I'd spin on in analysis paralysis. I can tell that people at my job are overusing cause they keep leaving Gemini's code comments in
I’m going to use it as slightly smarter UiPath for repetitive dog shit work and that’s it
I use AI for exactly one application in my life, writing quick-and-dirty code that i'm gonna use like once. and then there are people out there like "I asked chatgpt for ideas on what to make with the stuff in my fridge tonight
I use AI when I want a quick and dirty answer to a complicated question Google won’t answer without a wikipedia deep dive. It was really cool at first but it was also obvious you’re not learning anything along the way, unlike the epistemological katamari that a Wikipedia deep dive is.
“epistemological katamari that a Wikipedia deep dive is.” that is so perfect
It's a tool for life's equivalent of NP problems, and people are just going all in on using it for life's equivalent of P problems.
So like this is really not the worst use of AI
Taking a picture of your fridge and getting a recipe? That's like an actual improvement in living standards
Oh and I used it once last week to extract text from a PDF that wasn't copy/pasting correctly
yeah some of the OCR w/ AI cleanup tools are actually really good now
I use it for 2 things - to keep a running list of items so I don't have to manually log them in a spreadsheet and to do math that I don't know how to set up an Excel formula for (and always verify it before I just go with whatever it spits out)
The use of AI to process my email inbox has been a revelation. Yesterday I typed in "read that email my son's soccer coach sent and put each training date and match on my calendar, cc my work calendar, my wife, and my kid" and that saved me so much time.
And you used that time to….
what tool(s) do you use for this?
curious about this too - sounds like either a custom setup using Gmail API + LLM integration, or possibly something like Zapier with AI actions? most email automation tools I know require more structured triggers, but the natural language parsing described sounds quite sophisticated.
I use the Shortwave email service and their AI is one of the features that keeps me subscribed. It’s really good at natural language interactions - see below where I just asked it when I was in Paris
ah, Shortwave! makes sense - built around AI from the ground up rather than bolting it onto existing tools. the natural language email search is particularly clever - way more intuitive than remembering date ranges and keywords.
Shortwave email
I use AI as my last resort search engine when all the rest fail me or as a way to have an idea or argument I’m making reformulated to see if I spot any mistake in my thinking (something that I did previously, without AI)
I just don't use it. Ever. I like looking up the information and verifying it.
I met a professor last month, appeared to be in his early 40s, who volunteered to me at a conference dinner that he'd used ChatGPT to compose a condolence email to his *best friend* upon the death of the friend's father. I was dumbfounded. Not only that he'd do that, but that he'd tell me about it.
That NYT story this week about the ChatGPT-assisted suicide was really dark, but one of the bleakest aspects was how the kid had the AI write their suicide note.
I'm finding it more and more essential to tell people directly that they're braindead and they need to develop or redevelop critical thinking. People are rotting their brains with constantly scrolling and they need to hear it
That's not surprising to me... ...well, maybe the telling you part. Humans feel all the feelings, but we often struggle with words. Asking an unfeeling companion to help draft the words that I would want to say, if I only could, seems like a natural application.
...and without judgment! We are, after all, just a bunch of monkeys. Even native speakers struggle with language—I think this is one of many reasons people love pop music! I do think "help me write a respectful condolence letter that captures my feelings" is perfectly valid.
yeah like templates & hallmark cards have existed for a while, it's a bit unfeeling but like it's certainly not the worst ick ever
You wouldn't have used a Hallmark card to send your condolences to *your best friend* though? A casual acquaintance, a none too close colleague, sure — but the shock here is he used it for the death of his best friend's father (and that as someone whose job suggests some autonomous skill with words)
So without getting an argument on social media (😅)—for me, it's the thought that counts, and often that does include effort. There seems to be as assumption that buying a Hallmark card or using LLMs *must* be because the person is lazy—it very well could be! But I don't think this is a given.
That said, most of the TechBros pushing AI slop on the rest of us and trying to get us to offload our (and their) emotional labor onto LLMs are absolutely not doing so in good faith and should *not* be given benefit of doubt.
Like most things, a tool is good, a better tool is better, but you still need to put in the effort -- unless you just don't care about what comes out at the other end. But for something like this, I don't see an LLM as much different than a stack of Hallmark cards + a thesaurus.
What the fuck are you all on about. Struggle or not the effort is what matters. Not the act itself. Putting effort in to find the words is what fucking matters. You can feel as awful as possible, but unless you put the effort in to tell us it's not going to be you. A robot is a cop out.
I'd be so fucking insulted if someone told me they couldn't even be bothered to write a condolence letter to me that came from their heart but that they offloaded it onto their glorified secretary.
And that's absolutely valid!
If that's what they did, then yeah, you're right, that's awful. To me, it would still be a struggle, a process, that would likely take many hours. I've never used anything an LLM wrote without doing at least a half-dozen passes, and for something like this, it would take me all day, even with help.
What a great friend… damn
It really does feel to me like a zombie movie, except that instead of staggering toward you mumbling "Braaiins" the zombies cheerily tell you about all the ways they've cheerfully thrown away their humanity.
“…with a whimper.”
WHAT
Yeah. It's one of those record-scratch moments I'll remember for the rest of my life.
I can't believe he told you. I can only imagine it as a confession. First time I had to do this I was 13; my best friend's Dad passed suddenly. I vividly recall how awful it felt to not know what to say. I wanted to say things useful and true, but I didn't know how. That moment was important.
Part of this is that Bluesky as a text-based platform self selects for people with a higher than average facility with words, who of course scoff at the use of GenAI to create new writing. But GenAI is very good at handling drudgery and many people’s daily workflows involve a lot of drudgery
The way I see it is that GenAI is an above average writer, and that appeals to the majority of people who are not above average writers. Especially people with English as a second language or who don't speak English at all
at my old job they would frequently use ChatGPT for creative writing and I would always ask that they let me do it - they weren’t necessarily better, but I was (quant finance firm), but they were hesitant because of the time involved, like might as well get this right away
we've seen this at my work, personally
i used to write emails for everyone on my team
I used to write emails for my mom and her husband when I was in middle and high school 😂
It's very simple. "If you can write better than AI, you hate AI. If you write worse than AI, you love AI." And most people are morons.
It's insane the stuff I see people using it for at our job, much of it in violation of data privacy regs. I use it for parsing public data like citations. Even Copilot, the Simple Jack of LLMs, can handle a .ris file pretty well as long as you make it a .txt one first.
We have so many problems with it at my job for reasons I can’t get into and like, yeah, the policy just does not exist in any practical sense
Same. What I can say is that I walked by a conference room a few weeks ago and saw a screen with [CLIENT PROJECT] as the header of a ChatGPT response. Why do people do this?