@MaximeRivest a real use case of this interface you must implement is have an obsidian vault as a second brain for all that you think and write in this journal. You can have an llm pre clean the input to the vault and then make the vault vector searchable say with qmd and then retrieve smartly
AI evolution:
1.Autocomplete
2.Full code
3.Agent
4.Harness
5.Loops
My guess for what’s next would be “essence”. Think of it like desktop to get the distilled essence of a body of content. Can be called compression.
@waitbutwhy I think it is just a limit where if you try to observe something smaller ( it takes more energy to observe smaller things ) then the energy required to observe at that scale would create a black hole.
@rossium Instead of doing 100c50*2^100. I don’t know how you even do it without using a computer. You can just estimate p(tie) as 1/sqrt(n). And get to 45% for case of 100 and 49.5% for case of 10000 etc.
I am taking a 1:30am night cab in Singapore and I am worried because the driver is watching reels.
And then I look at the reel and it has Bertrand Russel speaking something in his ears. I am not sure I should be more worried or less.