|
Marketing arm is going apeshit over AI as what they want is a word-salad spinner. For non technical poo poo it's a godsend; marketers don't want to write copy for the 80th marketing push this month, so something that can scan a product page and spit out highlights for you saves so much time they're clamoring over each other to add it everywhere. Unlike nfts, there are some uses, most of the hype seems to be around dumb tech poo poo it's bad at doing (but makes clickbaity headlines) or the legality of training data for image generating AI.
|
# ? Nov 2, 2023 16:01 |
|
|
# ? May 30, 2024 05:29 |
|
Thanks Ants posted:Am I unfairly cynical about ~*AI*~? I was watching this and it just seems like a system to pad emails out by including things that the person you're messaging already knows, because they wrote the message. So you're saying after all our years of bitching and complaining, somebody took it upon themselves to finally solve the "this meeting could have been an email" problem.
|
# ? Nov 2, 2023 16:43 |
|
There will still be meetings because people have to justify their roles, and filling days with meetings is all they can do
|
# ? Nov 2, 2023 16:46 |
|
Ideally LLM stuff will eventually end up as a useful tool for actual artists and writers to use to save some labor. First though corporations will try to use it to completely replace as many jobs as possible and we'll get a bunch of tragedies/comedies out of it.
|
# ? Nov 2, 2023 16:51 |
|
There are a non-zero number of times I consult an LLM to write me a script and just copy that bitch over to fix up to my purposes a little.
|
# ? Nov 2, 2023 17:14 |
|
Even the biggest proponents of AI writing frame it as a great tool to write things for other people to read. Nobody actually wants to read it themselves, and the moment anyone realizes they're reading AI drivel, they immediately stop.
|
# ? Nov 2, 2023 17:51 |
|
Yeah I assume that plugins to detect if you're reading AI generated text will be really common and people will just skip past stuff generated in that way
|
# ? Nov 2, 2023 17:53 |
|
Thanks Ants posted:Yeah I assume that plugins to detect if you're reading AI generated text will be really common and people will just skip past stuff generated in that way At least currently none of the ones that purport to do that are actually any good, having both high false positive and false negative rates.
|
# ? Nov 2, 2023 18:58 |
|
Search engines are already becoming increasingly useless as the majority of the results are AI generated keyword filled nonsense. We're going to see a return of webrings.
|
# ? Nov 2, 2023 19:31 |
|
ponzicar posted:Search engines are already becoming increasingly useless as the majority of the results are AI generated keyword filled nonsense. We're going to see a return of webrings. That would be good, because usually all I see posted is "no ring".
|
# ? Nov 2, 2023 19:38 |
|
A single ring on the left hand
|
# ? Nov 2, 2023 20:15 |
|
Thanks Ants posted:A single ring on the left hand the token ring, if you will
|
# ? Nov 2, 2023 21:04 |
|
I used AI to write answers to the bullshit fluff questions on my annual self-eval last month. I highly suspect AI wrote those questions in the first place.
|
# ? Nov 2, 2023 21:29 |
|
Drink and Fight posted:I used AI to write answers to the bullshit fluff questions on my annual self-eval last month. I highly suspect AI wrote those questions in the first place. And this is how the world ends, not with a bang but an ouroboros of AI generated poo poo.
|
# ? Nov 3, 2023 00:52 |
SyNack Sassimov posted:And this is how the world ends, not with a bang but an ouroboros of AI generated poo poo. It’s taking the human element out of a process that has already produced the equivalent of a poo poo-filled filing cabinet. Gonna change the world and let multiple levels of people pretend they’re doing something that produces 0 value. Might be the absolute best case scenario for AI - letting people realize wasting your life with nonsense is better left to a deranged Markov bot i am a moron fucked around with this message at 18:50 on Nov 3, 2023 |
|
# ? Nov 3, 2023 18:48 |
|
Drink and Fight posted:I used AI to write answers to the bullshit fluff questions on my annual self-eval last month. I highly suspect AI wrote those questions in the first place. Evals are poo poo, anyway. Self-evals doubly so.
|
# ? Nov 3, 2023 18:53 |
|
Sywert of Thieves posted:LLMs were brought up in our company a few months ago, and one of my senior co-workers said he used it to explain a piece of code to him. This is such a bad idea that I had a million things to say about it uit, but I was unable to accurately convey my into words.
|
# ? Nov 5, 2023 06:40 |
|
It's a language model, not a code linter/analyzer. It knows nothing about writing computer code, it just heard people talk about it online and parrots the information. It's a language model.
|
# ? Nov 5, 2023 07:38 |
|
I mean, yes, that's technically true, but it has helped me (and I assume your co-worker, too) actually solve real problems, so it must be able to recognize the patterns of code we're throwing at it to get useful solutions to them. It would one thing if it's responses were non-sensical and didn't actually help, but I can say with certainty that at least in my cases, it has been able to produce meaningful responses to my questions about what a function is doing, to the point that I've been able to then move on to my original task and succeed where I would have otherwise continued to bang my head against my desk. If you don't want to use it for that purpose, that's your prerogative, but to think your senior co-worker is doing something actively wrong by finding a way to make use of a tool at his fingertips seems like reverse-boomer-level nonsense.
|
# ? Nov 6, 2023 01:09 |
|
Thanks Ants posted:Yeah I assume that plugins to detect if you're reading AI generated text will be really common and people will just skip past stuff generated in that way i already skip past it because it has no meaning or value much like my posts
|
# ? Nov 6, 2023 17:54 |
|
nexxai posted:I mean, yes, that's technically true, but it has helped me (and I assume your co-worker, too) actually solve real problems, so it must be able to recognize the patterns of code we're throwing at it to get useful solutions to them. It would one thing if it's responses were non-sensical and didn't actually help, but I can say with certainty that at least in my cases, it has been able to produce meaningful responses to my questions about what a function is doing, to the point that I've been able to then move on to my original task and succeed where I would have otherwise continued to bang my head against my desk. I, too, trust black-box autocorrect to explain code to me.
|
# ? Nov 6, 2023 19:20 |
|
I pasted the Fast inverse square root code (without the "//"), and it gave me a pretty good reply. I would be scratching my head for quite a few minutes if I came across this in our codebase without a ton of comments.
|
# ? Nov 6, 2023 20:14 |
|
That's not really a representative test, as there are probably plenty of textual comments on that or similar code on the internet that it could have learned to replicate. The real question is how well it handles the spaghetti code from your intern 10 years ago that only exists in an internal codebase.
|
# ? Nov 6, 2023 20:20 |
|
Entropist posted:That's not really a representative test, as there are probably plenty of textual comments on that or similar code on the internet that it could have learned to replicate. The real question is how well it handles the spaghetti code from your intern 10 years ago that only exists in an internal codebase.
|
# ? Nov 6, 2023 20:28 |
|
Yeah, it's not that difficult to get them to spit out some of their training data in certain cases, which would include all the prompts you feed it to explain your company's valuable code.
|
# ? Nov 7, 2023 00:21 |
|
Turambar posted:I pasted the Fast inverse square root code (without the "//"), and it gave me a pretty good reply. Now try it with a piece of code that's equally complicated, but hasn't been plastered all over the internet with tons of explanations including a whole-rear end wikipedia article.
|
# ? Nov 7, 2023 03:18 |
|
I couldn’t get Google’s version of the AI chat bot to stop outputting snippets of PowerShell that at first glance looked correct but were using variables that just didn’t exist in the cmdlets, and never had done.
|
# ? Nov 7, 2023 05:11 |
|
WEBPAGES LOADING SLOWLY problem is actually ads are failing to load install adblocker, problem solved It blows my mind every time i run across someone who doesn't have something blocking ads already
|
# ? Nov 7, 2023 16:04 |
|
Shugojin posted:WEBPAGES LOADING SLOWLY The school can't install an ad blocker on the provided chromebooks because there's something in the district policy that prevents it. I don't know what idiocy that may be, but it's annoying AF and really makes them less useful.
|
# ? Nov 7, 2023 22:10 |
|
Shugojin posted:It blows my mind every time i run across someone who doesn't have something blocking ads already I had to turn mine off to make youtube work, and oh my god the internet is a hellish nightmare of blinking bullshit and autoplaying nonsense without one.
|
# ? Nov 7, 2023 22:14 |
|
AlexDeGruven posted:The school can't install an ad blocker on the provided chromebooks because there's something in the district policy that prevents it. I don't know what idiocy that may be, but it's annoying AF and really makes them less useful. It is probably a group policy that disables all installation of extensions.
|
# ? Nov 7, 2023 22:17 |
|
Methylethylaldehyde posted:I had to turn mine off to make youtube work, and oh my god the internet is a hellish nightmare of blinking bullshit and autoplaying nonsense without one. Try logging out and/or switching to Firefox (which interestingly enough also allowed Twitter to load again, though whether that's a good thing or not is arguable). And yes, I have ublock origin and ghostery installed in Firefox, as well as running through VPN.
|
# ? Nov 7, 2023 22:52 |
|
I think I might be going mad, but is Outlook randomly setting text alignment to justified for anybody else? Messages look fine and left-aligned when I compose them (I think) and then when I hit send they end up justified somehow. Edit: The problem seems to be Outlook on desktop (Windows, not the poo poo new app) displaying all emails justified, if I look in OWA they are formatted as they were sent.
|
# ? Nov 7, 2023 22:55 |
|
Thanks Ants posted:I think I might be going mad, but is Outlook randomly setting text alignment to justified for anybody else? Messages look fine and left-aligned when I compose them (I think) and then when I hit send they end up justified somehow. Did Immersive Reader inadvertently get enabled? It's in the ribbon.
|
# ? Nov 7, 2023 23:12 |
|
God loving dammit it's Microsoft trying out some things to be helpful https://answers.microsoft.com/en-us/outlook_com/forum/all/how-to-stop-outlook-from-changing-the/c20a041a-bde4-49af-bc38-25ccafd840e3?page=5 quote:Sorry for the delay getting back. I followed up with the Outlook and Word Teams about the email justification changing. It turns out the changes we are seeing are mostly expected and not bugs. The teams are implementing a feature called Automated Typography. This feature applies read only changes to emails in the Outlook Reading Pane and Word’s Read Mode to make the content more impactful and easier to read. Please keep an open mind about the feature and give it a chance. The product team has gotten some good examples from customers where the end result of their emails looked good, more professional, and were easier to read. I will ask if I can show some of those or if they can be added in a blog article. Thank you all for the feedback.
|
# ? Nov 7, 2023 23:23 |
|
Methylethylaldehyde posted:I had to turn mine off to make youtube work, and oh my god the internet is a hellish nightmare of blinking bullshit and autoplaying nonsense without one. If you must, just make an exception for YouTube, don't turn it off globally.
|
# ? Nov 7, 2023 23:28 |
|
Thanks Ants posted:God loving dammit it's Microsoft trying out some things to be helpful gently caress outlook rendering. The loving thing wants to render emails like a loving word document because that's the only html parser they have apparently. Every other web inbox renders poo poo sanely, why can't you, outlook, you gently caress?! Now they're just straight up changing the message rendering on a whim because it looks nicer? loving hell.
|
# ? Nov 7, 2023 23:30 |
|
guppy posted:If you must, just make an exception for YouTube, don't turn it off globally. Oh I know, but for one glorious half hour while I was fighting with poo poo, it was off completely. And ho-lee-gently caress, I am never doing that again. Website breaks when adblocker is enabled? Better find a different website, apparently. I do like how the ad blocker arms race now includes 'please disable your ad blocker' popup blockers, and blockers for the blockers. It's a nightmare of javascript all the way down.
|
# ? Nov 7, 2023 23:33 |
|
dragonshardz posted:It is probably a group policy that disables all installation of extensions. We asked, because one of the sites they had to use during the pandemic shutdown was loving awful. It's a school district policy that even they can't install a blocker. We already can't install extensions.
|
# ? Nov 8, 2023 03:09 |
|
|
# ? May 30, 2024 05:29 |
|
Ticket for "Computer keeps making a chime noise". Tried to replicate the problem, none of the tones I produced in Windows matched reports. I think its new users getting a swarm of "new printer found" messages. I just muted the computers in question and closed the ticket lol
|
# ? Nov 8, 2023 21:45 |