Lightnews — Scholar-powered news

Part of the issue to me is intelligence can either mean intelligence in "smarts" sense, or intelligence as in intelligence gathering.

( Its why calling things smart phones is kind of misleading. )

S.R. Weaver ☑️ @lwflouisa.bsky.social · 3h

Now Artificial Intelligence ( with intelligence means intelligence gathering ) thats very much possible.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 3h

Self Modifying code has been out for some time, basically as long as weve had personal computers.

The amusing thing is how AI companies think thats a sign of intelligence. Which it could be if made mapped to Bayes routines.

Thats not how rule rewriting is used most of the time.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 3h

I do quite like the concept of a meta-card game where a card game is able to do metadata analyses of its own shuffled card layout.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 6h

Shocker, its not like Sora users are horrendous human beings.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 7h

Not Canadian although I was thinking of art schools in Quebec, I didnt think about Canadian game studios even though that seems obvious now.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 7h

There is a few differences, but its a start.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 7h

I guess Ill look into Shoji Meguro and Kikuta then.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 9h

Cinnamon apple and hushpuppy. :D

S.R. Weaver ☑️ @lwflouisa.bsky.social · 9h

social.growyourown.services/@FediTips/11...

Fedi.Tips (@[email protected])

social.growyourown.services

S.R. Weaver ☑️ @lwflouisa.bsky.social · 11h

Also by their nature decision trees are a lot more sensitive.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 11h

Here we go, the solution is using GeneralizedSAAD to do coordinated and competitive dynamics, and then have my predictive model find coincences, so its split into different processes.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 12h

I think I may have to implement economies of reward in a different manner, or maybe not have such economies directly impact confidence.

Otherwise it be like if the state of walstreet influenced how you feel about Sloppy Joes.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 12h

For some reason reward signal isnt working in this one routine.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 13h

Yea if it seems like programmers are irate, its because were constantly having LLMs complain about methodology without any sense of context for why things are done a certain way.

Then adamant their way is correct.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 13h

See the bigger issue of LLMs is they dont understand the context of why you keep the two probabilities independant.

You as the end user is suppose to multiply that and not the machine.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 13h

Yes that seems absurdly simple, and thats the point: you need to be able to understand the math to audit the AIs findings.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 14h

With x and y being randomly chosen structured datasets with three unrelated facts.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 14h

But because its so powerful I have to isolate each time the function is called to a specific method with limited write access.

I used plus here as there is a joint function key.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 14h

Also I see world war LLM is heating up.

Im thinking of focusing on Naive Bayes and Expert Systems used in tandum.

( Technically this serves the opposite purpose of Bayes)

p(x + y ) = p(x) + p(y)

S.R. Weaver ☑️ @lwflouisa.bsky.social · 14h

Psychology Im fine with, I have very much a problem with ones wedded to controversial disorders not everyone agrees even exists.

S.R. Weaver ☑️ @lwflouisa.bsky.social · 14h

Keeping with exact month and day is something Ive had trouble with.

My current talking alarm clock reminds me what day it is, what number, and what time.

But I can make a spreadsheet about the different ways brands of alarm clocks tell time. And make bilingual within certain limits.