participating in my first hackathon today
i’ve generally always led front facing positions, no professional engineering background
first time building and shipping something entirely by myself for anyone other than myself
excited for the challenge, good luck to the other contestants who most likely won’t see this 🫡
@etubruton@RichardHanania 100% agree. I feel like this problem, as well as others people have, tend to come down to user error.
You have to craft the perfect prompt to get the perfect output you’re searching for.
If it doesn’t do what you had pictured in your mind - it’s almost always user error.
I think no one’s base writing form is eloquent. Even the best writers text their friends in shorthand.
You have to choose to write “better” intentionally.
Many people give LLMs prompts that don’t direct it to dig deep.
Then they get mad when it stays at a surface level or gives something generic.
From my experience, LLMs can be phenomenal writers. Capable of invoking true emotion in humans. It’s all about the human directing it.
If your LLM gives you only slop writing - it’s probably user error.
@Noahpinion@signulll I feel like some sort of automated snapshot and data compression process looped could solve many memory issues at small/medium scale.
Obviously problems will still arrive the larger any set of data becomes - but us humans have gotten pretty good at solving scale.
I believe.
same base building blocks. same universe.
consciousness emerged once from arrangements of matter.
what makes you so sure it can only happen once, in one way, from one specific arrangement?
@leecronin The only reason we are even able to comprehend a possibility space is through our own probability space - which is entirely based off the past.
Why can’t agents, if not now, eventually work in the possibility space too?
the vast majority of AI dialogue I see in the space focuses too much on the scientific aspects alone.
the philosophical side of the conversation is equally if not more important imo
This sounds incredibly impressive. Thanks for sharing!
On the UI/UX issue, I’ve been working on some possible solutions regarding this for myself.
Have you tried creating a sort of “design manifesto” of what you’re looking for
to feed ChatGPT?
It’s done wonders for my output.
Scientists just copied a Fruit Fly's biological brain and trapped it inside of a computer.
Not an AI model trained to act like a fly... A total digital copy of a fly !! This is some sick sci-fi stuff:
- They scanned and copied the brain, neuron by neuron, synapse by synapse, from electron microscopy data.
- Then dropped that brain into a simulated body in a video game like environment.
The fly walked. It groomed. It fed. Nobody taught it anything. The behavior was already in the wiring.
The entire premise of modern AI is that intelligence is something you train into a system. This is proof it's something you can transfer out of one. Wild times
appreciate you sharing this. i’ve created something similar for myself.
really cool to see someone else doing it too.
i plan on making a version that is as modular, customizable, and dumbed down as possible with the purpose of sharing with friends/family.
would love to pick your brain
@alexbasil@solana_sailor@thedankoe love this. currently working on an official form of something similiar that can be easily replicated and used by others.
definitely think it's a heavily underexplored use case that creates insane personalized value.
@btcjammer@agentic_joe@CtrlAltDwayne Ok cool, thanks for clarifying!
I’ve never touched /model so assuming mine is still in default High state, if that’s even a feature of Sonnet or only reserved for Opus
@btcjammer@agentic_joe@CtrlAltDwayne Interesting. My plan throughout the weekend is actually to test exactly that idea.
Re-do all the work i’ve done the last few days, this time with Opus as the model instead of Sonnet, and then compare outputs.