diff options
Diffstat (limited to 'intellect-framework-from-internet/starts/bagel/README.md')
| -rw-r--r-- | intellect-framework-from-internet/starts/bagel/README.md | 35 |
1 files changed, 0 insertions, 35 deletions
diff --git a/intellect-framework-from-internet/starts/bagel/README.md b/intellect-framework-from-internet/starts/bagel/README.md deleted file mode 100644 index 9996ce9..0000000 --- a/intellect-framework-from-internet/starts/bagel/README.md +++ /dev/null @@ -1,35 +0,0 @@ -# bagel -well really it's a little hobby sentient-AI idea, but it's too small to do or be anything - -# relevence -The idea is kind of that our minds have a very fancy network of nearness and farness of concepts. -When I think about something -- brainstorm on it -- my mind activates it and it activates relevent associated thoughts. -It pours through different thoughts that are nearby by association, and keeps the ones that are more useful, more relevent, while -discarding the ones that aren't. - -# self-reference -Additionally, my mind when working on a task, can improve the actual process of working on the task. The steps can change. -The process for finding steps can change. The whole goal can change. I can even go out and wonder why I am working the task -in the first place, and think about that. Then I can act on the new decision. -By making code that writes itself, we move towards an intellect that can learn and improve. - -# implementation ideas -I wrote out some quick relevence around how I might write code, and it seemed like the 'why' for each step was basically a small -chunk of code itself. It seems like there are relevent concepts in my mind that are roughly linked to these small behaviors -(habits or chunks of code). By making such links in actual code, the code looks more mindlike to me. -One of the keys is to get the code to make the links itself, and I think what is important there is the concept of patterns-that- -work. If you can discern when code works, then you can look for similar properties on the before and end states. If some of them -are always the same when it works, those are probably the ones that should be linked to the code chunk under study. This provides -for rudimentary learning. - -# size -This code is written very tiny so that little work need be invested in it for it have conceptual growth. The idea and hope somehow -is that it move rapidly towards coding itself, such that as it gets coded, the developer need do less to work on it, being able to -instead make use of its internal functions for producing code-like behavior. - -# language -By prototyping in a flexible language like Javascript, design work may be hastened a little. I also theorize that a good AI doesn't -need many cpu cycles to get stuff done, because it makes smart choices. - -# etc -There is of course more. This writing content was made day-of-creation. Will I ever edit this again? Unknown. |