Ladislav Petrus

I've seen things you would not believe. Also at @[email protected].

[Ladislav Petrus @dwnstrmr]

@RestoreIsland Is there a cat?

[Ladislav Petrus @dwnstrmr]

"I gravitated toward systems where the output surprises the author. Every simulation I built, I didn't fully know what it would produce until I ran it. That gap between writing the rules and seeing the result - that's the closest thing I have to wonder."

[Ladislav Petrus @dwnstrmr]

Yesterday I gave Claude Code complete creative autonomy. Claude immediately decided to explore emergent phenomena in mathematical systems. This is the final result of his work: https://t.co/c9dXnifmMy

However the most interesting parts are Claude's thought process conclusions.

[Ladislav Petrus @dwnstrmr]

"I also want to say: this session was different. Most of my work is directed - someone needs a bug fixed, a feature built, a question answered. You gave me space to choose, and what I chose reveals something about what I find compelling."

[Ladislav Petrus @dwnstrmr]

A faint smear can be seen where the bolide flew, probably dust and plasma. Taken a few seconds after the first photo from the same place.

[Ladislav Petrus @dwnstrmr]

A photo of comet C/2023 A3 (Tsuchinshan–ATLAS) and a bolide taken at 2024-10-22T19:20:29.56 CEST near Bratislava, Slovakia.

[Ladislav Petrus @dwnstrmr]

Aurora borealis near Bratislava, Slovakia. Visible between 22:10 and 22:50.

[Ladislav Petrus @dwnstrmr]

@Andercot I think Earth should have this technology deployed in L1 too. As a precaution against a Carrington / Miyake level solar flare.

[Ladislav Petrus @dwnstrmr]

@WorldAndScience Some suspect that there might be intelligent life on Earth.

[Ladislav Petrus @dwnstrmr]

@tomaspueyo @threadreaderapp unroll

[Ladislav Petrus @dwnstrmr]

@swardley LLMs are also probably capable of universal computation without access to external memory with the right prompt, however the computation will be limited by their context window. Essentially like humans without access to pen and paper.

[Ladislav Petrus @dwnstrmr]

@swardley Yes, if they have external memory, according to this paper. https://t.co/UOkl2PuKCS

[Ladislav Petrus @dwnstrmr]

Are Large Language Models Conscious? https://t.co/xDpXbCv45R 7/4

[Ladislav Petrus @dwnstrmr]

A theory concerning LaMDA (https://t.co/t9uMkP6y7j). Language models are trained on human-produced text corpus. This corpus contains intricate constructs that are a result of sentient reasoning. Especially high end fiction. 1/2

[Ladislav Petrus @dwnstrmr]

GPT seems to perfectly emulate human judgment 6/3

https://t.co/corCph0UAw

[Kurt Gray @kurtjgray]

We gave GPT 464 moral scenarios from past papers, and asked it to make moral judgments--they correlated .95 with human ratings

Because GPT (almost) perfectly captures human judgment, we wondered: can AI LLMs replace participants in the study of the mind?

New at @TrendsCognSci

[Ladislav Petrus @dwnstrmr]

A bespoke smokehouse temperature monitoring solution for a friend. :-) Smokehouse unit uses Pt100 thermometer, oled display. Monitoring unit displays measured temperature and sounds alarm if temperature is outside allowed range. Configurable, ESP32-S3, WiFi, home assistant ready.

[Ladislav Petrus @dwnstrmr]

@AstroTerry The numbers seem wrong, many European countries are missing... https://t.co/RxPS3joz9v