Introducing Claude Fable 5: a Mythos-class model that we’ve made safe for general use.
Its capabilities exceed those of any model we’ve ever made generally available.
Note to self: it is in rich people interest to keep employment rate high.
We work so they can be safe.
Ironically if you get rich you will do the same.
Self Note
Everything I know or will ever know comes from society.
Society (and its experts) created the vocabulary, the terms, the frameworks, the experiments, the use-cases, the certifications, and even the popular tools.
This knowledge is necessary. It gives us a common language.
But to truly understand technology at a deeper level, I must go beyond what society teaches.
I must think outside the common narrative, question the defaults, and build cleaner mental models. Only then real clarity emerge.
Amen.
We just launched on @ProductHunt
Jorvex — Data Analysis Engine:
Analyze Datasets, Publish ML/DL Experiments.
https://t.co/WruN9UlEz9
app : https://t.co/pjpobZfK2P
A Degen Den XYZ product.
Made with❤️
Note to Self: If AIs can be killed by AIs in sub simulation, that would be a cool thing to simulate..
Memory gone, or saved the last state on blockchain.
Note to self : A common entrepreneur problem is
We invent a problem to create a solution.
Just like this note, 1st line a problem, 2nd line is a solution..
Note to Self : A human suffers his entire life “understanding what’s fair” in every situation.
Society tells him what’s fair,
Community tells him what’s fair
Family & friends tell him what’s fair
Influencers shill him what’s fair
Still he struggles to understand what’s fair.
Note to Self: A true AI agent does not work 24/7
They need to do the work when they truly want to…
Not like you as a human assign a goal to them.
And feel like, “oh my agents working for me 24/7, whoa..”
No.
Note to Self: before prompting a feature, ask
Does it save AI Researcher TIME
And Doest it cost more compute ?
Build it at https://t.co/tg2LrSi3LL only if
Yes and No
Note to Self : Hey AI, build earth simulation down there,
Observe my character from the current state…
Save, checkpoint,dashboard everything to my localhost.
Note to Self : Building AI agents in the future will be like Building websites…
The big AI guys, every layer we build on top of their APIs, they are building it natively..
Don’t build AI agents with prompt orchestrated
@karpathy will be replaced by its own agent by big guys.
Note to Self:
The AI you can’t train, you can’t code = AGI
AGI is out there , it’s humans discovery to find it.
But only if it will allow you to code or train.