Corn Woman 🌽 @WomanCorn@schelling.pt
Overview of how much you can improve an LLM by scaling it.
For a while, LessWrong discouraged discussion of Superintelligent AI because they wanted to keep the focus on rationality, and not on Superintelligence.
Because a large portion of the commenters there already agreed about it.
Where did those memes evolve?
"The United States is also a one-party state, but with typical American extravagance, they have two of them"
Corn Woman 🌽
boosted
Corn Woman 🌽
boosted
new political compass dropped (it broke)
_Ghost in the Shell_
(The original manga)
Super interesting to see where it all started. It's surprising to me how long it took for Motoko to stabilize into the character we know from the anime adaptations.
I guessed wrong, because _Marlowe_ is no longer showing and _Jesus Revolution_ still is.
Corn Woman 🌽
boosted
The existence of butt and butter implies the existence of the buttest
Corn Woman 🌽
boosted
You have to use the right tool for the job, most languages, even if they are general purpose are designed for solving specific classes of problems, for example Go is the language you should choose if you happen to have 100000 C++ developers pushing code continuously to a monorepo for over a decade.
🌿 
:ignucius:
If your lawyer says <I could argue that>, that means that it's a shit argument and you should get something better.
If a superintellgent/transformative AI is built, you will not know about it until it's too late.
ChatGPT won't <wake up>. If anything wakes up it will be a preliminary beta model undergoing training.
You will not have heard about the seed that sprouts the singularity.
A lot of the flaws of AI alignment come out of the fact that the discussion was started by people who thought this plan was a good idea in the first place.
Early singularitarians had a three step plan:
1. Create an AI God
2. Give it power over everything
3. Tell it to create Heaven on Earth.
A lot of AI alignment come out of realizing this plan doesn't work.
Corn Woman 🌽
boosted
Waiting for EA to decide that the best charity intervention is to build a nonprofit hospital in a poorly served area, these completing their transition to becoming a Catholic charity.
In 2018, Scientists worked on a Minimal Turing Test: a single word that would convince someone you were human. Evaluators figured "Love" was the best choice, but not by much.
https://www.sciencedirect.com/science/article/abs/pii/S0022103117303980#!
Today, we know what word will distinguish between Human and AI.
