Post B7OnYrNIEF4PWiD3QG by futurebird@sauropods.win
 (DIR) More posts by futurebird@sauropods.win
 (DIR) Post #B7On92veeEGnpeyOCu by IngaLovinde@embracing.space
       0 likes, 0 repeats
       
       @Goopadrew @futurebird it's even better than that, the defense was "it doesn't matter if the AI is wrong because _everybody knows_ AI answers are bullshit and not to be taken at face value"
       
 (DIR) Post #B7OnYqQnjiHYbIqLFg by futurebird@sauropods.win
       0 likes, 0 repeats
       
       A company or a person can sue Google for misinformation in their AI overview. But when the AI overview says that Lasius myops is a social parasite when this ant is clearly being confused with Lasius interjectus who can sue on her behalf?If it says that Lasius brunneus is "invasive" and so someone wipes out their colonies (even though they are a helpful native species) who can bring a case of defamation for the ants?Lies cause harm, but there is only a way to protect capital from this harm.
       
 (DIR) Post #B7OnYqiWfoSbUH8Usa by cbuddenhagen@mastodon.nz
       0 likes, 0 repeats
       
       @futurebird Hi I didn't get that claim from Google AI. The only reason I checked is because invasive implies both being outside its native range and harmful. Your point is a valid concern regardless because of the unreliable outputs from AI. Even invasive ants that are harmful outside their native range are clearly not easily described as such within their native range. If brown tree ants were to establish on an isolated island, for example, they could harm the native fauna. I've worked in several Pacific Islands where invasive species, including ants, are a huge deal. #IAS #invasivespecies
       
 (DIR) Post #B7OnYqvHuMfW7r6gls by futurebird@sauropods.win
       0 likes, 0 repeats
       
       Inspired by this outcome. https://www.yahoo.com/news/world/articles/court-finds-google-liable-false-155629224.html
       
 (DIR) Post #B7OnYqwhp5nqCFloyu by IngaLovinde@embracing.space
       0 likes, 0 repeats
       
       @cbuddenhagen @futurebird > I didn't get that claim from Google AI.> The only reason I checked is becauseAre you aware that even the same LLM can produce vastly different "answers" (continuations) even to the identical questions in the identical environments, depending on the internal state of random number generator?Are you aware that, even despite that, even if we had deterministic LLMs, the architecture of LLMs implies that "answers" even to the same question but phrased slightly differently don't have to be "consistent" with each other?
       
 (DIR) Post #B7OnYrNIEF4PWiD3QG by futurebird@sauropods.win
       0 likes, 0 repeats
       
       I feel just as harmed as this company by all of the disgusting AI ant slander I have encountered. I'm just as mad. MORE mad.
       
 (DIR) Post #B7PqMdHbI2Z9MpMsZk by cbuddenhagen@mastodon.nz
       0 likes, 0 repeats
       
       @IngaLovinde @futurebird Tested it once with colleagues asking about a recommended recipe using the same wording.   Two people with the same IP got the same answer, but on another IP, I got another. So I've seen it.