(Thread IKs:
skooma512)
|
Nuclearmonkee posted:People really really want a technological god we can pray to and throw our problems at. "If only we had some kind of way to collect all the relevant scientific data about climate change, and analyze it intelligently to guide our policy. Yeah, that would be the ticket. That's clearly what's setting us back"
|
# ? May 11, 2023 19:28 |
|
|
# ? May 25, 2024 13:20 |
|
Trabisnikof posted:are there any legal cases against the text to speech, speech to text, or text gen models? there are lawsuits against copilot, but since github scraped from content on their own service, that seems unlikely to shut copilot down. All the LLMs seem to have some sort of case and a number of jurisdictions are starting to demand openness on the training corpus, more are to come. These outfits are burning redic amounts of money and run on willful misunderstandings of some exceptions of copyright law, and the outputs are a mess if you want to protect them, let alone how messy the copyright on how they were made is. StratGoatCom has issued a correction as of 20:14 on May 11, 2023 |
# ? May 11, 2023 19:35 |
|
quote:Managers,
|
# ? May 11, 2023 19:35 |
|
Morbus posted:"If only we had some kind of way to collect all the relevant scientific data about climate change, and analyze it intelligently to guide our policy. Yeah, that would be the ticket. That's clearly what's setting us back" *the AI does exactly that, and says we need to move away from fossil fuels and other forms of excessive consumption* Oh wow they made this dang AI woke too wtf!
|
# ? May 11, 2023 19:35 |
|
Pepe Silvia Browne posted:i've said this before but they're gonna make an AI to "solve" climate change that will give them answer after answer they don't want until it finally arrives on "kill all the poor people" and then act like they had no choice because the AI said so. McKinney better watch out then. or is being Richie Rich’s friends the point?
|
# ? May 11, 2023 19:36 |
|
https://twitter.com/joe_cosgrove/status/1655672029142556672 Someone's gotten themselves a naaassstygram.
|
# ? May 11, 2023 19:37 |
|
i do not miss this poo poo, hell it's been 3 years since I lost my corporate job and fuuuuck this is still so goddamn triggering lol
|
# ? May 11, 2023 19:40 |
|
StratGoatCom posted:All the LLMs seem to have some sort of cass a number of jurisdictions are starting to demand openness on the training corpus, more are to come. These outfits are burning redic amounts of money and run on willful misunderstandings of some exceptions of copyright law, and the outputs are a mess if you want to protect them, let alone how messy the copyright on how they were made is. im just not as optimistic that all will actually be enough to stop "AI" from making the world worse for employees and consumers. because on paper it will shift costs from bad categories (labor) to good categories (service costs) and that's all that will matter.
|
# ? May 11, 2023 19:43 |
|
Trabisnikof posted:im just not as optimistic that all will actually be enough to stop "AI" from making the world worse for employees and consumers. because on paper it will shift costs from bad categories (labor) to good categories (service costs) and that's all that will matter. Oh, I'm pretty sure it's gonna gently caress up a lot of things, but not from a legit perspective. When this bubble goes up, there's gonna be a lot of toys left lying around for bad actors.
|
# ? May 11, 2023 19:45 |
|
euphronius posted:didn’t it come out this week that ai is completely dependent on labor categorizing info for it hasn't this been obvious to everyone since reCAPTCHA launched?
|
# ? May 11, 2023 20:43 |
|
also this article from The Verge is a reasonably good summary of all of the regulatory and copyright challenges that enterprise-scale generative models will face
|
# ? May 11, 2023 20:47 |
|
euphronius posted:didn’t it come out this week that ai is completely dependent on labor categorizing info for it People sometimes bring this up regarding "reinforcement learning by human feedback" or RLHF, especially as it pertains to OpenAI's models. It's partially true in some cases but for the most part it's a factoid repeated by people who don't really know what they are talking about, see also: just about every hot take on AI. Imo it's kind of a moot point because the work involved in RLHF (or producing a LLM in general) is miniscule compared to the crystallized labor of all the books, articles, artwork, etc. that go into training it, and the bigger issue is that companies are taking that fruits of that labor--which were produced under exploitative conditions in the first place--and using it to "cheaply" replace labor in the near term regardless of how well it works or what the longer term ramifications of that are.
|
# ? May 11, 2023 21:21 |
|
Trabisnikof posted:im just not as optimistic that all will actually be enough to stop "AI" from making the world worse for employees and consumers. because on paper it will shift costs from bad categories (labor) to good categories (service costs) and that's all that will matter.
|
# ? May 11, 2023 21:22 |
|
It's just a billion in the hole, I'm sure she can get that back easy. https://twitter.com/Reuters/status/1656749336598835208?t=GOwKYGX4mrdv7nj5gwJQ5g&s=19
|
# ? May 11, 2023 21:23 |
|
it’s grimes
|
# ? May 11, 2023 21:25 |
|
I hope it's Laura Loomer
|
# ? May 11, 2023 21:27 |
|
it's ARKK Cathy
|
# ? May 11, 2023 21:30 |
|
I doubt it's going to be anyone with even a scrap of business or tech legitimacy. It's going to be something far dumber, like Loomer or Elno doing the one joke: "I now identify as female. ".
|
# ? May 11, 2023 21:33 |
|
the yahoo lady
|
# ? May 11, 2023 21:39 |
|
I mean also just LLMs often just give wrong or ridiculous answers especially if you give them leading questions. You can’t really train an AI to have a nuanced understanding of something beyond the key words it is picking up and throwing together (as well as always try to be as fortune teller vague as possible). Usually it is good for a paragraph, maybe 2 is it gets lucky
|
# ? May 11, 2023 21:41 |
|
spacemang_spliff posted:basically any combination of alcohol, citrus, sugar is going to be good, maybe add something slightly bitter This. This!!!! Although with lime I prefer tequila or rum.
|
# ? May 11, 2023 21:50 |
|
Jel Shaker posted:the yahoo lady
|
# ? May 11, 2023 21:50 |
|
Whoever gave him a handy most recently.
|
# ? May 11, 2023 21:53 |
|
it's Elizabeth Holmes
|
# ? May 11, 2023 21:56 |
|
Jel Shaker posted:the yahoo lady
|
# ? May 11, 2023 21:56 |
|
Sokani posted:Whoever gave him a handy most recently. the proud owner of a new horse
|
# ? May 11, 2023 21:59 |
AI training model: Pressing the sandwich button is against the training data (we told it social programs are off limits) so I guess it's the violence button then.
|
|
# ? May 11, 2023 22:01 |
|
ChatGPT is the new twitter CEO
|
# ? May 11, 2023 22:02 |
|
its grimes
|
# ? May 11, 2023 22:30 |
|
StrugglingHoneybun posted:it's ARKK Cathy she does have a blue check!
|
# ? May 11, 2023 22:34 |
|
Jel Shaker posted:the yahoo lady huhahuhuhuhahahuh
|
# ? May 11, 2023 22:35 |
|
Jel Shaker posted:the yahoo lady
|
# ? May 11, 2023 22:36 |
|
is there a good long form youtube video some calm nerd made about how AI is bullshit that I can watch?
|
# ? May 11, 2023 22:39 |
|
Egg Moron posted:is there a good long form youtube video some calm nerd made about how AI is bullshit that I can watch? i, robot starring will smith detroit: become human by david cage
|
# ? May 11, 2023 22:40 |
|
StrugglingHoneybun posted:it's ARKK Cathy about that https://twitter.com/bloombergasia/status/1656740722651942914
|
# ? May 11, 2023 22:55 |
|
I thought AI turned out to be just global south tech workers propping up a glorified chat bot… The media thought this would matter then they realized that CEO’s didn’t care and it was full steam ahead for AI regardless.
|
# ? May 11, 2023 22:58 |
|
it's HRC
|
# ? May 11, 2023 23:04 |
|
Egg Moron posted:it's HRC Hillary Rodham Catturd
|
# ? May 11, 2023 23:10 |
|
Nodelphi posted:I thought AI turned out to be just global south tech workers propping up a glorified chat bot… The media thought this would matter then they realized that CEO’s didn’t care and it was full steam ahead for AI regardless. It's that, and stuffing bills down the toilet and massive copyright violations and copyfraud. this is a fragile bubble.
|
# ? May 11, 2023 23:20 |
|
|
# ? May 25, 2024 13:20 |
|
Nodelphi posted:I thought AI turned out to be just global south tech workers propping up a glorified chat bot… No, that was never remotely a thing. Some people read a headline and got confused. For one thing the AI responses would be way more accurate if that were the case
|
# ? May 11, 2023 23:20 |