after playing around with gpt-3 for a while, imo this analogy is the best way to interact with the tool.https://twitter.com/a_yawning_cat/status/1339826316074348544 …
Show this thread
-
a game like townscaper is a good example of how the WFC algorithm works. (which is similar to the WFC as a concept in quantum mechanics)https://store.steampowered.com/app/1291340/Townscaper/ …
Show this thread
with townscaper, you can see that "observations" around a building "collapse" or "condense" around the existing homes. the new condensate fits the "shape" of the existing architecture so as you flesh it out, it gradually grows into an entire village
-
-
you don't have authoritarian high modernist control over what the output is. you can only say "hmmm this looks interesting, let's add more here." the only thing you can control is where you focus your attention and what you choose to call forth into existence
Show this thread -
so for gpt-3, your prompt has to be something like a high dimensional vector pointing into a direction you find "interesting" however, this space is likely 99% empty even with the language corpus being the entire internet.
Show this thread -
this is similar to olber's paradox. even though there are many many many stars. if you just point in a random direction, you're more likely to hit darkness rather than light.https://en.wikipedia.org/wiki/Olbers%27_paradox …
Show this thread -
so imo, the ability to riff off of metaphors and analogies is gpt-3's biggest strength.https://twitter.com/a_yawning_cat/status/1340904815816179713 …
Show this thread -
i have found that some people very much dislike metaphors/analogies because they find them too "magical". however they can actually feel very "intuitive". they connect two concepts through an "arc" or "portal" that skips logical deterministic pathinghttps://twitter.com/a_yawning_cat/status/1341959274964996096 …
Show this thread -
this means that a metaphor/analogy "Points" from a Here to a There that is definitely THERE. this means that once you find one metaphor that successfully delivers a payload & connects the dots, you can use gpt-3 to figure out how to deliver other payloads along the same arc!
Show this thread -
so classes of usage like "use it like a chat bot" work by incrementally refining and growing a pointer. it's actually an erroneous model to ascribe it agency when it's more like a knee jerk reaction searching for closure
Show this thread -
with twitter threads, there a sensation of "open" and "closed" just like with stories there is a clear sensation when the story is "finished". (just like a joke "closes" w/ a punchline) the story tapers and there is a sense of finality and closure at a final "point"
Show this thread -
rather than get gpt-3 to actually DO something specific, a better tool might be to find some way to manage these different pointers, cutting off low resonant directions but feeding interesting intermediary results into a main branch
Show this thread -
similar to how the "shape" of a blockchain points in the "direction of god" by requiring hashes to be prefixed with zeroes, this tool would help gradually refine the shape of a initial seed thought by helping to twist and "point" it in the right directionhttps://twitter.com/a_yawning_cat/status/1405674449571962883 …
Show this thread -
e.g. an example is a thread of ideas that you wish there were more of, or one that ended too quickly. or a thought that you can lightly grasp but have trouble cohering it into something more resonant
Show this thread -
it helps explore the feeling of "knowing" that something is "there" and "i'll know it when i see it" and allows u to work it out rather than bump around randomly for it e.g. like growing a random plant until the first flower blooms to see its "final form"
Show this thread
End of conversation
New conversation -
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.