Yonae
Nowhere Girl"The use case for LLMs is spam." Spam email, spam websites, spam news stories, spam books. Nothing anybody wants to read. Software that costs a million dollars a day to produce shit.
The guys peddling this software don't know the difference between spam, "content," and actual creative work. It's all spam to them already, or they would like it to be because they can't do it. They undervalue creative work because they don't value anything. They just want to be handed a pile of money for turning in shit.
Kalshann@gwynnion
This, just all of this!
This, just all of this!
Jarjan (He/Him)@gwynnion
Yep... The cost (time, money) and skill required to create shit has just gone down to near-zero with #ai #llm while the cost to create something worthwhile stays the same. As long as they can sell shit, the business model will be to sell shit.
Another reason to abolish (post liberal) capitalism.
Yep... The cost (time, money) and skill required to create shit has just gone down to near-zero with #ai #llm while the cost to create something worthwhile stays the same. As long as they can sell shit, the business model will be to sell shit.
Another reason to abolish (post liberal) capitalism.
Keith@gwynnion yup, and the marketing industry as a whole gobbles it up. I'm working on an MBA. In my digital marketing class, we're being taught how to use generative AI (chatGPT and its ilk) to generate marketing copy because "it's the future of the industry."
Personally, I think marketing copy has always been drivel, but the crap spit out by chatGPT is worse than most. Yet when I take the 💩 chatGPT gives me, spend maybe 5 min polishing it a bit and turn it in, the marketing prof says its great.
Personally, I think marketing copy has always been drivel, but the crap spit out by chatGPT is worse than most. Yet when I take the 💩 chatGPT gives me, spend maybe 5 min polishing it a bit and turn it in, the marketing prof says its great.
FediThing 
Pete Alex Harris🦡🕸️🌲/∞🪐∫@petealexharris @keydelk @gwynnion
Oh dear. What a world this is becoming. 😞
@FediThing @gwynnion senior marketing executives have no qualms about burning the career ladder - they’ve already climbed to the top. In the past, executives would ask junior marketers to write the copy for a marketing campaign they came up with. Now they ask chatGPT to write the copy and don’t have to hire junior marketers.
Jason St-Cyr 
@keydelk @gwynnion I work in the digital marketing software industry. We are under constant pressure to deliver more AI integration into our tools because all marketing teams want to type something like "Title for an article about pandas being extinct" and have it spit out the marketing-style copy. It does marketing copy VERY well and current search engines value 'new' content and repeated content creation on a topic. That forces a "content crisis" as some call it, so in come the generators.
@jasonstcyr @gwynnion exactly! Which is why so much of the web is filled with computer generated cruft that pleases Google, but sucks if you’re trying to find good information.
Kit Walker@gwynnion i remember first realizing this when Clarkesworld had to close submissions because they were getting inundated with LLM spam.
someone had clearly, at some point, made a get-rich-quick guide that positioned short fiction mags as reverse vending machines--insert wordcount, receive cash. and a whole bunch of other people believed that person, because they didn't read short fiction (or much of anything else).
Iñaki Calvo 🇪🇺@gwynnion not just spam, but an awful lot of it, for sure.
Tor Iver Wilhelmsen@gwynnion not just spam but sometimes danger; though I haven’t seen any myself, I have seen it reported that in the stream of AI-written books shoved onto the Kindle store via Kindle Direct Publishing are mushroom guides that say poisonous shrooms are safe. (In addition to the cases of books pretending to be written by established authors.)
Violet MadderI DON'T LIKE SPAM
Nini@gwynnion A perfect use case for scammers, grifters and bad actors.
Simon Willison@gwynnion I've been an almost daily user of LLMs for over a year and I promise I'm not using them for spam
I use them as a thesaurus, a coding assistant, (weirdly) a brainstorming partner - they are a rubber duck that talks back to me
I think generating text for publication is one of the least interesting applications if the technology, though I get why professional writers tend to assume that's their main use
And maybe it is! They're incredible spam machines for people who want to do that
🌚JR🌝@gwynnion It helps me with coding scripts at work. Has saved me hours by giving me workable code. I just have to edit it.
Juan Per¢ent,🇲🇽 🍉,🇻🇪@gwynnion Do not forget content farms for video in YouTube. So called “ Cash-cow” channels produce millions of hours of mindless AI content, sprinkled with some human intervention…bigG makes profits from advertising as would be entrepreneurs shoulder the production costs to fill the channels. All spam
Børge@gwynnion Don't forget misinformation and propaganda for the rich and powerful
LLM is perfect for this, instant generation of text etc supporting whatever line you want to push.
Trump's adviser Steve Bannon said their aim was "to fill the media with shit" in order to prevent proper oversight of the government, and LLM is tailor-made for this task.
Chris (Master of Potate) 🥔@gwynnion I like it as little code helper to shave of some busy work from programming. But mustly it is just a toy. It is great for all the places that stopped producing good things to read and replaced it be word vomit with SEO tricks. Or spam as you called it.
Poligofsky@gwynnion I think the best positive use case is translation, including writing descriptions of images/video, and transcriptions. Such tasks can be tedious for people, but easy to verify after the fact, so we can feel pretty comfortable trusting machines to do it.
Analyzing images for patterns that defy human recognition can be put under this umbrella, too.
But, yes, confabulation, deception, and "flooding the zone" are dark timeline applications.
Time for regulation!
Mark Gardner
David Wheeler@mjgardner @gwynnion It’s just like winning the lottery
Dave Dawkins (D. Harrigon)@gwynnion Bloody Vikings.
Matthew Abbott@gwynnion LLMs also match the manufacturing process of spam quite well!
J.H.Noyes@gwynnion@mastodon.
I love LLMs, or Anthropic's Claude at least. I bounce ideas off of it and use it to zero on subjects to research. I understand there are limitations and that the devices are inseparably prone to confabulation, but they work well for me.
That said, they can also be used to produce mountains of convincing-sounding garbage, and that there are serious IP-facing issues with the garbage they create.
@robotistry
Foucault_this@gwynnion Solid general agreement, but a hard disagree at code generation for co-creating webapps and widgets for bespoke / personal / entrepreneur efforts.
Matthias
Richard W. Woodley ELBOWS UP 🇨🇦🌹🚴♂️📷 🗺️
Graydon@gwynnion I'd argue that making a comms channel useless by drowning it in shit is the point.
Already happened with traditional news outlets; now the money is coming for the internet. "Knowledge is power and nobody else should have any", more or less.
AlgoCompSynth by znmeb