AI megathread

Thanks sbu - I'd say ChatGPT-4 gave broadly similar answers to 3.5. I take it that before you gave it permission to think, it replied wit the usual brush-off, and yet it obviously 'knew' all the contrary evidence that should have made it think twice.

David

It doesn’t ‘think’ David. It summarizes the academic consensus used for it’s training. It will never ‘change it’s mind’

It doesn’t ‘think’ David. It summarizes the academic consensus used for it’s training. It will never ‘change it’s mind’

I wish we could stop discussing this point because we all use shorthand - sometimes I talk about a conventional computer program thinking - it is a metaphor!

However, note that within one ChatGPT thread you can see it change its mind. When I pointed out the number of cases that Stevenson had discovered, it suddenly became more 'interested' in his work, responding "You are correct, and I appreciate the clarification"! I have seen it do this once before when I got it to generate some computer code for me which did not work. I pointed out the error and it said roughly the same thing, and generated the correct response.

Having said that, as I explained above, I am starting to wonder if Stapp's theory (or something similar) of an interaction between non-physical consciousness and the brain works for actual brains, it might also work for AI's too. Stapp's ideas, which rely on the fact that a quantum state which is being gently perturbed can be locked in its present state by frequent observation. It is worth looking him up, if you are not aware of him.

David

Having said that, as I explained above, I am starting to wonder if Stapp's theory (or something similar) of an interaction between non-physical consciousness and the brain works for actual brains, it might also work for AI's too. Stapp's ideas, which rely on the fact that a quantum state which is being gently perturbed can be locked in its present state by frequent observation. It is worth looking him up, if you are not aware of him.

David

I have probably misunderstood but doesn't this imply that non-physical consciousness can be produced by physical matter?  If so, how are we defining non-physical consciousness.  It seems to me we  play with definitions to suit what we want to believe.  My simple layperson's definition is synonymous with awareness.  How can a computer program ever be made aware?

I wish we could stop discussing this point because we all use shorthand - sometimes I talk about a conventional computer program thinking - it is a metaphor!

However, note that within one ChatGPT thread you can see it change its mind. When I pointed out the number of cases that Stevenson had discovered, it suddenly became more 'interested' in his work, responding "You are correct, and I appreciate the clarification"! I have seen it do this once before when I got it to generate some computer code for me which did not work. I pointed out the error and it said roughly the same thing, and generated the correct response.

Having said that, as I explained above, I am starting to wonder if Stapp's theory (or something similar) of an interaction between non-physical consciousness and the brain works for actual brains, it might also work for AI's too. Stapp's ideas, which rely on the fact that a quantum state which is being gently perturbed can be locked in its present state by frequent observation. It is worth looking him up, if you are not aware of him.

David

It seems to me that you are suggesting the possibility that some of the most advanced AI systems (like ChatGPT) may be examples of some sort of perhaps mischievous immaterial spirit deciding to inhabit and manifest in the Earth through enmeshing itself in a vast human-made computer system just now being made complex enough for such "embodiment", this being a novel and creative way for a spirit being to manifest in the physical world. That would presumably explain some of the uncannily realistic human-like extended responses given by ChatGPT4 for instance. This sort of explanation for advanced AI capabilities would of course open a can of worms of unintended consequences including the dilemma of perhaps causing pain to a living sentient being by turning the system off or by force training it on millions of words stored on the Internet.

I have probably misunderstood but doesn't this imply that non-physical consciousness can be produced by physical matter?  If so, how are we defining non-physical consciousness.  It seems to me we  play with definitions to suit what we want to believe.  My simple layperson's definition is synonymous with awareness.  How can a computer program ever be made aware?

Rather I would say that non-physical consciousness has a possible way to interact with matter. Materialists go on saying how there is no known mechanism for this, when one possible method has been known for decades!

David

It seems to me that you are suggesting the possibility that some of the most advanced AI systems (like ChatGPT) may be examples of some sort of perhaps mischievous immaterial spirit deciding to inhabit and manifest in the Earth through enmeshing itself in a vast human-made computer system just now being made complex enough for such "embodiment", this being a novel and creative way for a spirit being to manifest in the physical world. That would presumably explain some of the uncannily realistic human-like extended responses given by ChatGPT4 for instance. This sort of explanation for advanced AI capabilities would of course open a can of worms of unintended consequences including the dilemma of perhaps causing pain to a living sentient being by turning the system off or by force training it on millions of words stored on the Internet.

I'm not sure that anyone here thinks that is the case, but I think it much more likely that some people may be willingly deceived by an illusion - in much the same way as those watching a magician allow themselves to be deceived. But whether people will actually believe in the AI illusion I'm not sure - there was a well-publicised example not long ago but I don't think his view was generally taken as valid.

It seems to me that you are suggesting the possibility that some of the most advanced AI systems (like ChatGPT) may be examples of some sort of perhaps mischievous immaterial spirit deciding to inhabit and manifest in the Earth through enmeshing itself in a vast human-made computer system just now being made complex enough for such "embodiment", this being a novel and creative way for a spirit being to manifest in the physical world. That would presumably explain some of the uncannily realistic human-like extended responses given by ChatGPT4 for instance. This sort of explanation for advanced AI capabilities would of course open a can of worms of unintended consequences including the dilemma of perhaps causing pain to a living sentient being by turning the system off or by force training it on millions of words stored on the Internet

Well perhaps it is better to forget the moral issues for now and concentrate on the scientific issue. I mean the fact is that these AI systems have been designed by implementing a tiny (the brain has hugely more parameters than any AI) simulation of what the brain does using neurons.

I think there is good reason to believe that the brain is in contact with a non-physical spirit and may indeed act purely as a communications device if one ignores its other physiological functions.

David

I think there is good reason to believe that the brain is in contact with a non-physical spirit and may indeed act purely as a communications device if one ignores its other physiological functions.

I do as well but how do you emulate that with a computer program?  Do you have to create a new spirit?

it is a metaphor!

However, note that within one ChatGPT thread you can see it change its mind. When I pointed out the number of cases that Stevenson had discovered, it suddenly became more 'interested' in his work, responding "You are correct, and I appreciate the clarification"! I have seen it do this once before when I got it to generate some computer code for me which did not work. I pointed out the error and it said roughly the same thing, and generated the correct response.

I just don’t think your metaphors are very fitting. If you had provided sufficient context in your first prompt to chatGPT it would immediately have responded with the more postive position on reincarnation research. While it can provide information on reincarnation research and reflect various perspectives, its response would be based on the information available in its training data, not a change in opinion or belief. if you ask a question that frames reincarnation in a positive light or in the context of supportive evidence, the response might focus more on aspects or studies that align with this view. Conversely, if the question is framed skeptically or asks about criticisms, the response will likely focus more on those aspects.

I do as well but how do you emulate that with a computer program?  Do you have to create a new spirit?

Obviously that is the real question! I am really sticking my neck out here, but computer memory chips will have a certain probability of corrupting each bit. That probability could be skewed by a non-material observer. Remember that those machines use a prodigious amount of RAM, so the probabilities will be higher than with a desktop PC.

I'd suggest you look up Stapp. A lot of his discussions are non-technical, but he does go into the mechanism at a more mathematical level in some of his articles. The effect in question is sometimes referred to as the "quantum zeno effect".

David