
Interesting, if there are #AI-powered robots trained on this paper and to which degree they may generalize if you ask them to help...

Working on Qubic, Aigarth, and Paracosm now. discord.gg/2vDMR8m. Speaks Assembler, BASIC, C, C#, HTML, Java, JavaScript, Pascal, Python, SQL.

Interesting, if there are #AI-powered robots trained on this paper and to which degree they may generalize if you ask them to help...

It's interesting how an advantageous trait of natural intelligence (x.com/SamaHoole/status/21218…) may be disadvantageous in other conditions (en.wikipedia.org/wiki/Judas_…). It hints that #AI successful in some conditions may fail in some others.
In 2004 the sheep on a Yorkshire moor worked out how to cross a cattle grid, and the village below them has not had a safe flowerbed since. A cattle grid is the simplest idea in farming. A pit across the road with steel bars over it. A hoof cannot grip the bars, so the animal will not cross. It has kept sheep on hills for the best part of a century. The hill sheep above Marsden, near Huddersfield, looked at theirs for a while. Then one of them lay down on its side at the edge of the grid and rolled. Over and over, eight feet across the bars, like a soldier going under wire. It got up on the far side and walked into the village. Others watched and did the same. Within a season the trick had gone through the flock. Villagers woke to sheep on their lawns, in their vegetable plots and browsing the churchyard. A local councillor told the BBC, with some feeling, that they were also clearing five-foot fences and squeezing through eight-inch gaps. Nobody had taught them. One animal solved the problem and the rest learned it by watching, which is how sheep learn everything, and why a flock knows its own hill. For as long as there have been jokes about sheep, the joke has been that they follow each other. At Marsden they followed the one that had cracked it, across a barrier designed by engineers for the sole purpose of stopping them.


@grok seems to have confirmed that LLM's abductive reasoning lacks *insight* - x.com/i/grok/share/c2hhcmQtM…. Maximum they are able to do is to recombine phenomena they know about. On the other hand, #Qubic's #AI is tabula rasa, #BPP9000 knows nothing about the universe.


As promised we just released @_Qubic_ #OpenScience #Neuraxon For @arcprize #ARCAGI3 dataset of MultiNeuraxon 2.0 evolved brains, 155,192🧠 604,045 Files 100Gb Months evolving and improving over 114+ NAS runs for the 25 Public ArcAGI3 games. #NoLLMs #NoGPU just bio-inspired evolved brains, By @josesanchezhb & @VivancosDavid Thanks to @c___f___b #Aigarth and the #Qubic community! @huggingface Dataset here: huggingface.co/datasets/Davi… @github github.com/DavidVivancos/Neu… More info, code, and demos soon!

How to test if your favorite "AI" possesses any intelligence: x.com/i/grok/share/610f0cecf…. @grok has failed.

ELI5 of @OpenAI's 722 papers drama: Alice claimed that 4+2-1-2=6. Bob calculated only 4+2-1 fragment yet. LLM simplified to 4-1=6 and disproved. Bob is unhappy that calculators have become more powerful and he no longer needs to spend that much time on mechanical work.

An #AI without access to a true random number generator (one that nobody can rig) is cooked. Bookmark this tweet. Within a decade, you’ll see I was right.

Have you ever wonder what place the humans will likely take in a world ruled by #AI?.. Right, scavengers, like the ancestors. Scavengers feeding on digital leftovers. Can we change this fate?..




Hey $KAS holders. Let's play a game, I've started longing your coin, are you joining? Hit retweet if you are or do nothing if you are chickening out.

We have identified interesting features of #Aigarth's #AI training algo: 1. It creates hypotheses, verifies them, drops wrong ones, returns to put aside ones; 2. It uses the same pattern for different phenomena if they have similarities, applies little modifications to the pattern if necessary; 3. It chains simple calculations into complex ones, rearranges them. All this looks similar to how hoomans reason.

Due to lack of participants and $KAS heading down I've closed the short with near-zero profit. We'll try to do it later again.