Post B94qhjnBk7uMBo5Gi0 by ariadne@social.treehouse.systems
 (DIR) More posts by ariadne@social.treehouse.systems
 (DIR) Post #B94n46P3xPags0CyRc by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       the thing about AI art is that the user creating it typically tries way too hard, to the point where people can just say "lol this is clearly AI"
       
 (DIR) Post #B94nDIibXTuFffjd2W by ariadne@social.treehouse.systems
       1 likes, 0 repeats
       
       good art tells a good story, when you paste a selfie into gemini and say "make me look like a superhero in a comic", you're not actually telling a story of any depth, just "hi, i'm bob and i aspire to be superman"
       
 (DIR) Post #B94nZaNSZ29WoITFx2 by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       which is great if you want to make clip art of yourself as a superhero in a comic, i guessbut what of that superhero?  what are his hopes and dreams?  how has he been heartbroken in the past?  where is the tension?the viewer can't ascertain the backstory, so it feels cheap
       
 (DIR) Post #B94nu6DVtXzrLnbH2u by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       i suspect a user who spent more time crafting their prompt could add context that produces a result with more depthbut that kinda defeats the marketed purpose of image diffusion: low effort visual elements to fill spaces where visual elements are expected
       
 (DIR) Post #B94oJjD63WorWecsVM by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       it is the same way with these AI music models like sunoeven if you apply a low-pass filter to delete the weird harmonics those models produce, the songs themselves are basically soul-less marketing jingles because there is no real story and no emotional tension
       
 (DIR) Post #B94oW9LYht6o3sVVb6 by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @dysfun it absolutely was shit
       
 (DIR) Post #B94omk5rqmyGfC50V6 by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       the underlying point being that largely these things could likely be overcome with better prompting, but the people who would know *what* to prompt are more likely to just go make the thing themselves
       
 (DIR) Post #B94pHNh5BM5mAX62ee by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       and this is the thing about language as the only mechanism of I/O to these generative AI services, frequently it is easier to just do a thing rather than describe the thing that should be donethere is a whole rabbit hole that we could go down about whether or not people actually think in terms of language (the tl;dr answer is that it is context-dependent, and the cognitive mode usually stays in sync with the active task), but that leads to other rabbit holes like "what even is cognition?" and "how do multiple streams of cognition collapse into observed first-person consciousness?"
       
 (DIR) Post #B94pln6nFtgzg8cgvw by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       the tl;dr answer to those rabbit holes are "we're not really sure, but we've mapped a lot of foundational neural networks"
       
 (DIR) Post #B94qF2rxjhfyxRdSfA by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       this makes stuff like #strudel really interesting by the way because using strudel is a cross-domain task: you're writing code that generates musicso, when using strudel, are you thinking in terms of music?  or in terms of language (specifically javascript)?
       
 (DIR) Post #B94qhjnBk7uMBo5Gi0 by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       the answer, incidentally, is really interesting: what winds up happening is that over time, neural connections between both task-specific networks get strengthenedso, you start to think in terms of both at once, and how one impacts the other
       
 (DIR) Post #B94qkDf1fEsnP5RvMm by tknarr@mstdn.social
       0 likes, 0 repeats
       
       @ariadne This. All roo often I see the C# I need to solve a problem. To apply Copilot the way work wants me to, I have to translate that back into English. That's 10x the work dealing with the imprecision and ambiguity of English when spelling out what I want done. We invented programming languages to eliminate that and make it easier to tell the computer what to do. WTAF?
       
 (DIR) Post #B94r1O04TXvYOTTktE by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @tknarr i think about code in terms of sequence points
       
 (DIR) Post #B94sFSKLHiJ5p1EXEO by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       there's a lot of caveats, of course (i am simplifying the concepts to make them more accessible), but essentially, over time, with continuous practice, both task-specific networks can begin to cooperate with each other and recruit each other for said cooperation
       
 (DIR) Post #B94skmv0mBXcf7KF0K by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       like there is a TON of literature using functional MRI data to study network recruitment in cross-domain tasks if you want to go down the rabbit hole
       
 (DIR) Post #B94t6sccFgrAinFtXE by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       fMRI incidentally is really neat
       
 (DIR) Post #B94tHluhX9Od6LuMwC by kouhai@social.treehouse.systems
       0 likes, 0 repeats
       
       @ariadne this makes me regret not doing cognitive science as a minor
       
 (DIR) Post #B95KfQeMwK0krmm32P by ariadne@social.treehouse.systems
       0 likes, 0 repeats
       
       @lobo yep its hard at first, but with practice you'll probably get there