You can, for example, give one an instruction to do something. After it has done it, you ask the other (verbal) hemisphere why they just did that, and they immediately make up a plausible-sounding explanation despite the scientists knowing it’s not the actual reason.
There’s an uncomfortable amount of similarities between how the human brain and LLMs work.
Next word correlating with the words said before, confidently spreading false information, telling people what they want to hear… These are by no means unique to LLMs.
Sounds like AI got something right then.
There’s an uncomfortable amount of similarities between how the human brain and LLMs work.
Next word correlating with the words said before, confidently spreading false information, telling people what they want to hear… These are by no means unique to LLMs.