4 ms·
I'm no expert on LLMs, but I've always found the "it's just predicting the next word in a sentence" trope to be a huge misrepresentation of what LLMs are actual
by mindwok 3y ago
I'm no expert on LLMs, but I've always found the "it's just predicting the next word in a sentence" trope to be a huge misrepresentation of what LLMs are actually doing. To me this is a really cool example of how LLMs actually learn more than just simple word prediction and start to exhibit real intelligence for some tasks.
- wddkcs 3y agoIntelligence, or consciousness? As a novice in all the technical aspects, I fail to see the distinction between a 'self updating internal model', and a human conscious experience. I think some of the current characterizations of LLMs is due to unwillingness to posit algorithms as entities.
- sr-latch 3y agoIs a PID loop conscious then?