One thought on “New fastbook, “Humans Are LLMs” …”
Hi Michael, I have done a lot of thinking about AI, LLMs, and thinking in the recent past. As a former professional computer programmer I notice the following: When an AI produces something that -if produced by a human – would be deemed a creative act, “Creatives” are quick to defend their craft as having some undefinable quality that cannot be attributed to a machine, or it’s output. But I suspect that if asked in a blind test, many people could not tell. I cannot recall where or when I’ve read this, but it is a recurrent theme in recent scientific and IT journals. It is understandable that”creatives” defend their output, because it’s their livelihood.
I’m already convinced that our brains probably do much the same thing as LLMs in a lot of cases. After all, Language is probably one of the greatest technologies ever invented.
I have also had a long standing beef with the term “AI”, because what we call “AI” really is not intelligent in the sense that you and I mean. I re-define it to be “Artificial Idiot”. But after reading “Humans are LLMs” I notice also that what passes for thinking for most people can- quite accurately- also be called “Artificial Intelligence”.
I believe that we (all of us) need to break free of our enchantment with AI, because researchers are already noticing and recording evidence of a cyclical ‘dumbing down’ effect. AI is used to produce web content. AI is trained on web content.
Bonus: check out recent episode of South Park where they mercilessly lampoon the sycophantic character of many AIs.
Hi Michael, I have done a lot of thinking about AI, LLMs, and thinking in the recent past. As a former professional computer programmer I notice the following: When an AI produces something that -if produced by a human – would be deemed a creative act, “Creatives” are quick to defend their craft as having some undefinable quality that cannot be attributed to a machine, or it’s output. But I suspect that if asked in a blind test, many people could not tell. I cannot recall where or when I’ve read this, but it is a recurrent theme in recent scientific and IT journals. It is understandable that”creatives” defend their output, because it’s their livelihood.
I’m already convinced that our brains probably do much the same thing as LLMs in a lot of cases. After all, Language is probably one of the greatest technologies ever invented.
I have also had a long standing beef with the term “AI”, because what we call “AI” really is not intelligent in the sense that you and I mean. I re-define it to be “Artificial Idiot”. But after reading “Humans are LLMs” I notice also that what passes for thinking for most people can- quite accurately- also be called “Artificial Intelligence”.
I believe that we (all of us) need to break free of our enchantment with AI, because researchers are already noticing and recording evidence of a cyclical ‘dumbing down’ effect. AI is used to produce web content. AI is trained on web content.
Bonus: check out recent episode of South Park where they mercilessly lampoon the sycophantic character of many AIs.