4
« on: October 06, 2024, 09:14:52 AM »
Hi everyone, I have a few insights you might be interested in since I used the paid version of gpt, all of them, to test how MLM 'learn', how they generate replies etc for one of my jobs.
First, the AI designation is a misnomer, a manipulative marketing misnomer that reinforces people's misperception of what constitute machine 'learning'.
That being said, when it comes to astrology, and pretty much every other subject, MLM work based on the curated literature they are fed, and it being curated is an important aspect, so the quality of the literature and the censorship of the same are key to what the Model will generate. This when it comes to astrology, and many other subjects like psychology, sociology, much of political science, etc is very very bad news. Since astrology is filled with junk as a subject. You have to understand that the feed in process of 'literature' includes any and all 'articles', blog posts, etc any self described expert has ever written on the subject. And almost all of the literature it was fed is 'modern astrology', some of it is 'vedic' astrology, and very very little, statistically almost insignificant, is the higher end of the literature or higher end commentary on the subject.
Since the Model uses statistical models to generate replies, it can't generate a higher quality than what the vast majority of 'literature' is on the subject. This is why it will be almost useless if you try to use it for Hellenistic astrology. It will always go back to 'modern', 'psychological' and 'esoteric' astrology, with some 'vedic' components here and there.
MLM also have an other weakness that most people don't know about or if they do they don't know the incredible impact it has. They are 'regulated' and heavily censored with what their developers call safety and 'ethical' guidelines, but first, these guidelines are terribly paternalistic and patronizing, even in the paid version, and they are seemingly meant for people that are either 6 years old or suffer from serious psychiatric disorders and thus need to be protected, and second, these guidelines interfere with the 'reasoning' capacity of the Model. The Model will ignore any and all information that triggers his warning system, and it will suppress, alter, etc any input that its guidelines react to, this results in 'cognitive' paradoxes where the generated reply is factually wrong simply because the guidelines are there to 'protect' you from facts. So, for example, it will not engage with Hellenistic texts (even if you try to create your own Hellenistic astrology gpt by uploading Hellenistic era text files and your own guidelines into a 'separate' bot) because its guidelines do not find them PC enough.
It will only generate, as a rule, only modern astrology platitudes, heavy on untestable general 'maybe, perhaps, sometimes' etc. and if by any chance you force it to abide by your rules by adapting your rules so they don't trigger its warning system, then it will generate what you feed it, but it won't be able to 'reason' beyond it.
So no, you can't use it, and I highly doubt that you ever will be able to use it, to 'read' a chart, beyond the limits I described, for not only will these limitations persist, they are getting worse and worse with each new generation. In order for a MLM to be able to do what I guess many here would like you would have to have a model that you yourself train, you curate the literature, you set all the guidelines, etc and you have access to the higher end computation capacity which even the paying models don't give you.
23 01' 40" 
00 45' 18" 
03 20' 39" 
27 41' 48" 
22 12' 11" 
21 15' 52" 
r
13 29' 11" 
r
26 29' 05" 
r
27 51' 12" 
29 38' 44" 
05 34' 44" 
21 09' 29" 
Home
Help
Search
Login
Register
Show Posts