ai (llm) thoughts (Patreon)

i posted two [instagram] stories yesterday: 

1) 'i am begging you to play with ai and understand what it can do for you....it has changed the way i learn completely...my whole life my information mind has felt like a droopy plant and now it's got a stake and it is growing'

2) '[ai] is a calculator but for everything'

i think i surprised people and i want to give more context. i understand not everyone is on board with even talking about ai and what it can do because they've relegated it to the moral atrocity dungeon of their working minds. i think it's good that the critics exist because they raise points of consideration that are deeply relevant and will change the way these systems are shaped. but i need you to understand: the critics will not stop ai from existing. this is here forever. it is a societal hinge...a before and after....an unstoppable force, like the internet.

i think people sense its inevitability and it is making them afraid.

before i talk about learning i want to address some of ai's common criticisms and my (very) brief thoughts, so you understand that my openness to it doesn't come from a place of thoughtlessness, or naivety:

environmental impactvalid. i think people should continue to push this. historically tech advancements come with clunky environmental trade-offs. the industrial revolution...power grids...early computing....steep costs. their value outpaced our ability to predict and mitigate impact. we recognized how essential these technologies were and we adapted. renewable energy. chip efficiency. it happens over and over again. that doesn’t mean adaptation is inevitable or easy but it is possible

arti have so much to say but like. look. people have used all kinds of tools to make all kinds of bad art for ages. it happens over and over again. digital cameras, machines that facilitated digital art, machines that made electronic music. we complained that those too would be used to flood the culture with garbage and for that reason argued they should not exist. but they are tools. ai is not an oracle, or a magic 8-ball, or a meaning vending machine, it is a tool. if you ask an llm for a poem and press 'go' you will get a shitty poem, just like how not everyone with a dslr is creating photography but taking pictures, just like how you can't throw a rake into your yard and expect to find the leaves in a pile (don't rake your leaves, by the way. it's better). more tools mean more bad art. novelty means at the moment the bad art is everywhere. i understand how this feeds cultural anxieties. i also think it’s important we come up with some kind of guidelines for credit and usage. but instead of pissing ourselves off because joe schmo changed his bio to 'ai artist' after birthing a harem of unclothed anime girls we can use this moment as an opportunity to think about what makes art meaningful to us and where we allocate our attention. say it with me: an opportunity!

skill losscalculators became widespread in the eighties. similar fears about skill loss came about. (i think they're comparable in that they are both tools which require discernment to use; can be misused or incorrectly applied to problems; skill augmenters that 'think for you'.) but calculators, !!!when used thoughtfully!!! (and !!!in the appropriate setting!!!), instead let us dedicate more time to this thing 'higher-order thinking' and complex problem solving. in the same way will ai solidify the foundations of our individual information minds and the one of our collective consciousness. i believe this sincerely.

there are others. basically: i think ai's potential for good outweighs all of its detractions, all of it, by orders of magnitude. it is a consciousness expander. it is a curiosity engine. it is super exciting.

all that being said. i posted those stories yesterday because: i started reading gödel, escher, bach: an eternal golden braid which was kindly recommended by someone in my last reading wrap-up...you're a superstar...it was exactly the challenge i needed....and it is a challenge! i'll talk about the book later. what you need to know for now is that it's math heavy. my math knowledge is, like, fine. i was moving slowly through the early sections and thought ai could help.

i want to talk about metaphors. metaphors are powerful conduits of understanding. metaphors make you go a-ha! because you don't need to dig entirely new paths; you're integrating information into existing schemas. look: llms are very good at coming up with metaphors. they can come up with a swath of them super quick. they can come up with metaphors that are personalized to you and your existing areas of knowledge.

here's what i do: i tell gpt 'hello!' (i always say hello) 'i am a writer, and a poet, and i spend a lot of time on the internet. i know a lot about these things. i'm also interested in nature and communication. with those things in mind, can you help me understand these concepts through metaphor?' i take a photo of the page. i feed it to the robot. i am served metaphors. i understand the concepts better than i did before. then i can individuate the original concept. where do these metaphors start to fall apart? what else do i need to know?

anyone who decries llms as a kind of glorified google is not understanding just how efficient, immersive, and dynamic they are, or how they allow for interplay between many concepts instantly. do you see? it's like inventing multiplication but for concepts. it's given me the confidence to take on more challenging material without fear of getting stuck or reaching a dead end in comprehension. it is a dynamic teacher who is with you all the time. it is an extension of you, like a phone or prosthetic, but for your learning mind. when interfaced with thoughtfully, it is a map through complexity. i basically intuited my way through high school and college without studying once and now for the first time i feel like i understand what 'learning' is.

here are some screenshots from our many-hour-long conversation:

can you tell me more? i'm interested in the idea that consciousness is related to an experience of time–can you suggest reading material to do with that idea? can you explain this to me like an amateur? like an expert? how would you represent this idea as a painting? what are the criticisms? what are the criticisms of the criticisms? what are some common misconceptions about this topic? can you quiz me? connect this scientific concept to an artistic concept. what's the best way to break down this complex topic into parts? how could i visualize this concept to understand it better? what strategies will help me integrate this information? what questions should i be asking to understand this more deeply? (you can ask it.....a question.....about what questions you should be asking!!)

i feel like i'll spend the whole day regulating my nervous system if i reply to comments on this one but please know i say all of this with an acknowledgement that change brings discomfort....with an understanding i might feel differently if my livelihood were directly at stake....with enormous sympathy for the teachers left to bear the brunt....with deep concern about the continued mistreatment of the planet which protects and feeds us......and with a total and ultimate love for art, creation, and all things human. i want only what's best for us. i sincerely think this is part of it!!

sav xo

edit 1: the people are saying claude is better! try claude

edit 2: changed use of 'ai' to 'llms' in a couple places

edit 3: got a little high and made humor edits