Every single time I think about what LLM are I think about this quote from the game Night in the Woods:
“We’re good at drawing lines through the spaces between stars like we’re pattern-finders, and we’ll find patterns and we like really put our hearts and minds into it and even if we don’t mean to.”
LLMs are based on neural networks. They are little brains that have nothing else to deal with than finding patterns in our own logic and can seem to be smarter than what they really are because of it. Evolution has not weighted them with an ego or urgency, but because it has been trained on ours it can sometimes emulate it. But it fundamentally lacks the complexity of our brains, at least for now. It is still amazing what they can do given so little, and it is amazing how convincing they can be with their answers when they are completely wrong. It is a viral form of intelligence.
LLMs are actually just massively improved spell checks. If you’ve used an IDE with in line error detection, technically that’s AI now.
I do wish we’ve drawn the line more clearly on what “AI” usage means in terms of “this game was made with AI”
Every single time I think about what LLM are I think about this quote from the game Night in the Woods:
“We’re good at drawing lines through the spaces between stars like we’re pattern-finders, and we’ll find patterns and we like really put our hearts and minds into it and even if we don’t mean to.”
LLMs are based on neural networks. They are little brains that have nothing else to deal with than finding patterns in our own logic and can seem to be smarter than what they really are because of it. Evolution has not weighted them with an ego or urgency, but because it has been trained on ours it can sometimes emulate it. But it fundamentally lacks the complexity of our brains, at least for now. It is still amazing what they can do given so little, and it is amazing how convincing they can be with their answers when they are completely wrong. It is a viral form of intelligence.