Dawkins declared himself unable to determine consciousness through the chat terminal, which is the reason the Turing test is relevant.
Try imagining the same 'gotchas' in the original Turing test, (i.e. you're told beforehand you're talking to an AI, and you have insider knowledge of how AI works.) Then the role of the test-taker is to simply disregard the chat and to already know the answer.
Dawkin's posts might be gross and out of touch, but let's at least get a proper rebuttal - what definition of consciousness, when applied to interactive chat, could differentiate a person from an LLM?
I used to think conscious was just being able to say "nah not doing that", but that's typically based on rules, and those rules can be programmed in always.
I don't know, I made a comment a week or so ago asking the same thing, why is our neural network conscious? We're also very easy to poison
Since ELIZA, the more we try to apply the Turing Test, the more evidence we get that the Turing Test about human gullibility, rather than saying anything meaningful about the machine [1,2]. Surprisingly it turns out that even critical thinker Richard Dawkins is quite gullible.
As we get older we seem to lose some critical thinking. We become more afraid. Celebrities fading from the spotlight often become desperate for attention. My guess is these factors are motivating Dawkins to grasp for any foothold as he slips into obscurity.
A younger and less famous Dawkins may have been more skeptical of Claude.
The dedication to create a domain name with dear x y to explain to a dimwit that a next token predictor is not conscious is s level of misery that no sentient being should ever descend into.
Dawkins declared himself unable to determine consciousness through the chat terminal, which is the reason the Turing test is relevant.
Try imagining the same 'gotchas' in the original Turing test, (i.e. you're told beforehand you're talking to an AI, and you have insider knowledge of how AI works.) Then the role of the test-taker is to simply disregard the chat and to already know the answer.
Dawkin's posts might be gross and out of touch, but let's at least get a proper rebuttal - what definition of consciousness, when applied to interactive chat, could differentiate a person from an LLM?
I used to think conscious was just being able to say "nah not doing that", but that's typically based on rules, and those rules can be programmed in always.
I don't know, I made a comment a week or so ago asking the same thing, why is our neural network conscious? We're also very easy to poison
You cant? But so what
Since ELIZA, the more we try to apply the Turing Test, the more evidence we get that the Turing Test about human gullibility, rather than saying anything meaningful about the machine [1,2]. Surprisingly it turns out that even critical thinker Richard Dawkins is quite gullible.
[1] https://medium.com/@innovariart/the-turing-test-no-longer-me...
[2] https://arxiv.org/abs/2405.08007
As we get older we seem to lose some critical thinking. We become more afraid. Celebrities fading from the spotlight often become desperate for attention. My guess is these factors are motivating Dawkins to grasp for any foothold as he slips into obscurity.
A younger and less famous Dawkins may have been more skeptical of Claude.
The dedication to create a domain name with dear x y to explain to a dimwit that a next token predictor is not conscious is s level of misery that no sentient being should ever descend into.
They probably got their AI Agent to knock it out in five minutes.