With new studies coming out showing that GAI is a long long way off because the models collapse rapidly with growing complexity, things are going to get weird for those who thought this was actual true thought. Too many smart people are being absolutely buffaloed that this is real cognition rather than an impossibly complex, lightning quick variation of an if/then/else algorithm capable of new levels of data display.
You know what, I'm going to make a post about this. You got me ranting and 3 paragraphs in I'm just scratching the surface. Watch for it next Monday. I need content for my weekly "musing".
Part of the reason genAI for images worked was the curation of the data was extensive. There are still issues, but they are more minor and less obvious. The issue with LLM were we are trying to make a knowledge database without curation, without verification. We are just putting in everything. It is the old garbage in, garage out.
With new studies coming out showing that GAI is a long long way off because the models collapse rapidly with growing complexity, things are going to get weird for those who thought this was actual true thought. Too many smart people are being absolutely buffaloed that this is real cognition rather than an impossibly complex, lightning quick variation of an if/then/else algorithm capable of new levels of data display.
You know what, I'm going to make a post about this. You got me ranting and 3 paragraphs in I'm just scratching the surface. Watch for it next Monday. I need content for my weekly "musing".
GIGO
Garbage in garbage out!
Part of the reason genAI for images worked was the curation of the data was extensive. There are still issues, but they are more minor and less obvious. The issue with LLM were we are trying to make a knowledge database without curation, without verification. We are just putting in everything. It is the old garbage in, garage out.