(01-13-2025, 06:04 AM)19Bones79 Wrote: I decided to ask a question of Elon's computer. 

It doesn't prove anything but I thought it was fun. 

...

I get where you're coming from with your questions to Grok, but I don't really have a lot of confidence in the responses of AI at present.  AI is nothing more than a product of information available on the interwebz.  When you consider how much control the MSM and social media have over the internet, you can really only expect to get confirmation bias as an output to questions like this.

That's just my opinion, but it seems to hold true from what I've seen so far.  I mean just think about it...if you put a computer inside a Faraday cage and don't connect it to anything, and then teach it a vocabulary of (100) words.  Then ask that computer a question.  The output you're going to get is some combination of those (100) words and nothing more.  If you ask it a 2nd question, you're going to get a response which is a combination of the (100) initial words plus the words added by the 1st question...and so on.  AI at present is just a data harvesting tool, with some backend automation to assemble that data in a coherent fashion.

Now, obviously there's a tremendous amount of information available for harvesting on the interwebz, but here's the problem.  Unless the AI knows how to interpret this information and put it into context it can't really utilize it in a meaningful way.  So, take a medical text, or a scientific white paper, for example; there are assumptions that are made in these documents about the level of 'understanding' possessed by the reader.  Unless the AI agent has that same level of understanding then it can't assimilate that data as part of its intelligence.  I'm sure developers will cross this bridge eventually, but the tools aren't mature enough yet.  Consequently, while much of the AI generated data looks impressive, it's really a victim of the old GIGO algorithm for the time being.

Just my .02