this post was submitted on 21 Apr 2024
30 points (100.0% liked)
TechTakes
1490 readers
36 users here now
Big brain tech dude got yet another clueless take over at HackerNews etc? Here's the place to vent. Orange site, VC foolishness, all welcome.
This is not debate club. Unless it’s amusing debate.
For actually-good tech, you want our NotAwfulTech community
founded 2 years ago
MODERATORS
you are viewing a single comment's thread
view the rest of the comments
view the rest of the comments
tired: stealing hundreds of dollars of electricity to mine hundreds of pennies in crypto
wired: spiking some project manager’s OpenAI bill to unsustainable levels by having their chatbot generate the worst nonsense ever experienced by a human
it feels a little mean doing this to a library, even if their use of AI is obviously doomed to failure, so a list of public access GPT prompts would be a service.
Yeah, hospitals/libraries/schools/etc should not be things on such a list generally
(In two minds about some of the US colleges, but that’s a different kettle of barbs)
inspired: crowdsourced prompt-based captcha solving
As soon as some of these LLMs get a math module to do math correctly (And not just via the LLM lookuptable thing) people could write scripts to externalize some more intensive calculations needed for crypto mining. Sure it will be inefficient as fuck, and I doubt the chance of getting a coin reward will be low, but it will be free.
last week there were a couple of articles about how easy it is to craft an input that makes public chatgpt bots execute scripts (usually as root) on their hosting containers, which is almost definitely the result of a module like that being implemented for better programming-related results (aka fucking cheating), so this is very likely already happening
Happy to at least not be the first to think of that idea, and sad to hear people will wreck the commons more.
WHAAAT
links plz?
found the original post! https://mastodon.social/@kennwhite/112290497758846218 the prompt to make them execute code is incredibly basic. no idea right now if the exploit is in the chatbot framework or the model itself though
oh shit, somehow I figured you knew already! I’ll skim through my browser history and masto boosts and see if I can find one of the articles
counterpoint: it gives openai more money
not necessarily/could be offset? openai is still in that "we'll set fire to money to make ourselves look good" stage of VC dreamery; find entities operating on credits, slap there
but possibly even in the case where it's still straight transactional, it might be a net negative for them: revenue, actual usage, and still no meaningful shift on their product becoming good. it'll just make them look even worse
the bigger problem (to which this suggestion would most certainly contribute things getting worse) is that they're still burning other important resources. I don't really have a good/clever proposal to this which isn't something like "well, burn their DCs to the fucking ground" (or other more creative forms of invasive service interruption)
they will burn through that money pretty quickly and without turning profit, however this can contribute to one of their ratfucked metrics to go up, which could hype up some segment of stonk market. because now they're fueled by hype and vc money, any new thing that would sustain that hype would be a bad thing (adoption here, kinda, at least as seen through excel)
i agree, 120mm mortar is much cheaper, faster, more irreversible, but openai going bankrupt and forced to sell their kit at least would generate less waste
and a net positive in terms of human happiness