BF2S Forums AI is going to kill us all.

complaining about generative AI slop overtaking the creative industries is not the same thing at all as being a luddite …?

your brain seems to have gone soft lately. are you feeling alright? like an apple with a bite out of it, slowly oxidising.

i mean, using it to troubleshoot something or fetch you some basic information in a trice does not seem like a problematic use-case. search engines were already doing just that, just with a few extra steps (and without the creepily ingratiating, pally tone of LLM chatbots).

feeding it the collective works of 100,000 authors without credit or remuneration and then employing it to write entire books, yes pretty problematic actually.

Last edited by uziq (2025-05-28 07:58:09)

if i recall correctly, either openAI or claude or one or another of the 'major' models in use today trialled and then reverted a version of their chatbot that was insidiously, egregiously keen to please the user. it was practically emotionally manipulative. people complained en masse because it was just making them uncomfortable to interact with. that is fucking crazy.

lol

Well I'm holding onto my CDs.

Says the person still using twitter

Bit of a stretch to call me active.

I think AI has very quickly moved into just about everything. It is more entrenched into our daily lives than we recognize.

...

I completed a curriculum project last summer. Some high school students used it as a jump off point for their submission to the National History Day competition.

I used AI to summarize NYT articles, produce intro paragraphs, and edit out watermarks from old photos for my curriculum project. The kids straight up submitted the AI modified photos to the history competition. The people who ran the competition would have no idea the photos were edited using AI. They wouldn't know that some of the intros they used were made up by AI after I gave the broad details to ChatGPT.

It is wild that the AI has seeped into so many things and did it very quickly. I can only imagine how many things around us are at least AI modified.

that's eerily the exact vibe i got. i was about to make some snipe about dupont and c8 and all the corporate perfidy.

i think sometimes it's like wwe fans going "ooooh!" whenever a fake elbow strike whooshes by a guy's head and sends him sprawling onto the mat. they know it's fake but suspend disbelief and get into it.

where i draw one of many lines on shitty youtube "documentaries" is when someone tries to engage me like what they just watched was all fact. bro, one of the people in the background had 3 legs and 5 feet for a few frames. this is not real footage.

fox doesn't even need to dig through archives of arson and bloody riots to falsly portray a demonstration in seattle now. they can just gen it. i'm already starting to see a lot of chatgpt-looking turns of the phrase in various news articles and blogs. i don't want to call it out because i don't want to be that asshole if the person wrote something legitimately and it just looks ai-styled. we're going to have generations of writers informed by the wheedling tone of chatbots, and i don't look forward to it.

Dilbert_X wrote:

Fake it till you unicorn? Builder.ai’s Natasha was never AI – just 700 Indian coders behind the curtain

https://techfundingnews.com/fake-it-til … e-curtain/

this is one of the reasons i'm quite sanguine about how much, or at least how fast, the AI revolution is going to affect most SMEs. as long as the major players in silicon valley are constructing their moats and asking high prices (and as long as chinese competitors like deepseek don't utterly crush that business model) – for now most regular businesses with modest budgets are going to be in this sort of bodge-job wild west of hastily tacked together AI solutions for problems that barely existed.

ca. 10 years ago in publishing the genius business move was to offshore as much of the technical infrastructure as possible to 'tech hubs' in places like chennai, those sorts of places where there's 10,000 coders for every street food vendor. now the entire industry is basically stuck with these service providers who struggle to put together a lean, efficient, and bug-free WYSIWYG proofreading applet. no way are they ever going to implement high-quality AI. not for the next 5 years at least.

i distinctly remember saying to my publishing director at the time, "wait, so the genius idea is to leave behind adobe acrobat and let indians design a worse version of a PDF mark-up app? that's entirely online and shared between editors and authors, like google docs? from a company that has 100th the talent pool and 10,000th the funds of google or adobe?" what could possibly go wrong? project managers are dipshits.

Last edited by uziq (2025-06-04 04:23:40)

it's a problem intrinsic to lots of the large LLMs. they seem to get worse over time. some sort of self-referential over-training phenomenon (don't ask me; i don't care enough to know).

google's enshittification has been proceeding apace for years. they don't care that their half-arsed google implementation returns pure slop. it hasn't hurt their advertising bottom line, or the other areas of the business where they've been pointed in recent years.

i find it genuinely troubling to see how often that top portion of the search page returns arrant nonsense. it rather calls into doubt every single fact or figure it produces above every search query now. we're effectively going backwards, from a library-search-style query and a trawling through results/archives for the right answer, to a vibes-based 'summary sounds right to me' reality. nevermind that the stats it produces are hallucinated.

and yeah, most businesses do not need AI. unfortunately there's normally a middle-tier of project managers whose entire existence can only be justified by trying to continually onboard new shit or subscribe to spurious workplace cod-philosophies like 'six sigma kaizen board just-in-time continuous improvement'. and they love a new bauble. who cares if AI doesn't work? that's their performance review bonuses sealed-in for the next 2-3 years. busy busy!

it's an irony of history that we could have probably had LLM tech in the 1980s, but for the fact that cybernetics research and that branch of tech was considered tainted by its association with communism and central planning. the cybernetics guys who would have been working on these problems were all considered pinko megalomaniacs. the neural networks underpinning it all date back to the cold war.

Last edited by uziq (2025-06-04 04:52:26)