I think an AI could be programmed to have doubt. It would probably be a worthwhile addition.
I think what you are calling doubt is the ability to weigh up the evidence and decide which interpretation is the best/most useful/most likely.
So far, from what I have seen, the ChatGPT takes a 'sunny side of the street' interpretation of the data that it has available. I think this is because it's corpus includes lots of opinion pieces and corporate puffery (my interpretation). I have found, though, that if you put back to it a question that challenges the positive, it will give you what it has from the other perspective. It does not (seem to) have the built-in ability to look at your question and interpret it from several different angles, then give you the different angles and an opinion about which is likely the most accurate, looking at (say) who was the source of each perspective and what they might have been hoping to gain by having their opinions prevail.
I agree with you that it will be difficult to program that ability into an AI. It is 'judgement', though, rather than doubt.
My vote is Empathy as a uniquely human skill -the ability to understand and share the feelings of others. I would cast doubt on doubt. When I think about doubt from a machine's perspective, I think about confidence level as a measure of doubt. If confidence level in the data or results is high, then doubt is low. If confidence level is low, then doubt is high and the machine may broadcast this and "ask" humans for help (or to intervene).
In the spirit of openness I must confess I'm not a voracious reader as our fearless leader is. I'm a thinker, relying on perceptions and relationships I've cultivated. I use my imagination to envision a better world and work to make it happen. I fail miserably, but am hopeful.
Many great thoughts on this topic. There are many attributes of humans that make us unpredictable and visionaries. I doubt if an AI could write a story like "Father Sergius" by Tolstoy. Computers may never be able to dream, although current research on dreaming as memory sorting may mimic such an experience.
I would say that doubt is a part of the inductive reasoning process, which AI can learn to process. I reckon that the human skill that AI won't be able to emulate - at least not to the depth & extent that some humans have it - is 'inventiveness'.. I could be wrong though, but my understanding of AI is that it has to 'learn' from humans; and inventiveness is the one thing that one can only learn to a limited extent. It appears that we all have it to a greater or lesser degree. Consider the life of a farmer: he has to 'make a plan' at least 3 times a day on average, improvising and urgently solving new problems like a rusty irrigation valve breaks and he has to 'make a plan' real quick; a fire-break must be urgently burned at exactly the right strip of grass for the current wind direction; his borehole pump breaks and the local supplier quotes him 2 weeks delivery on a replacement and he has to 'do something' that will supply water for the next 2 weeks... it's that "make-a-plan" with a piece of wire and a used handkerchief and the lid of the wife's washing-basket factor that I don't believe AI will ever do as well as humans.. I may be wrong though
AI can run routines to question its conclusions, but I think the doubt you are talking about varies depending on the human's experience. AI can mimic experience, but not fully incorporate a life long human experience. AI currently, has its limit, whereas human doubt changes over time, based on an individual's experience with others and the world. FAITH is my answer.
Judgement. Humans are able to make ethical and moral decisions, in ways AI will never be able to do.
Truth. Large language model like ChatGPT predicts what words go together based on input text and all the material it has been trained on. It doesn't care if the information it serves up is true or factual in any sense of the word.
I think an AI could be programmed to have doubt. It would probably be a worthwhile addition.
I think what you are calling doubt is the ability to weigh up the evidence and decide which interpretation is the best/most useful/most likely.
So far, from what I have seen, the ChatGPT takes a 'sunny side of the street' interpretation of the data that it has available. I think this is because it's corpus includes lots of opinion pieces and corporate puffery (my interpretation). I have found, though, that if you put back to it a question that challenges the positive, it will give you what it has from the other perspective. It does not (seem to) have the built-in ability to look at your question and interpret it from several different angles, then give you the different angles and an opinion about which is likely the most accurate, looking at (say) who was the source of each perspective and what they might have been hoping to gain by having their opinions prevail.
I agree with you that it will be difficult to program that ability into an AI. It is 'judgement', though, rather than doubt.
My vote is Empathy as a uniquely human skill -the ability to understand and share the feelings of others. I would cast doubt on doubt. When I think about doubt from a machine's perspective, I think about confidence level as a measure of doubt. If confidence level in the data or results is high, then doubt is low. If confidence level is low, then doubt is high and the machine may broadcast this and "ask" humans for help (or to intervene).
Real and utter vulnerability. In a way, connected to doubt as if you do not doubt yourself, you would never open up / be vulnerable.
In the spirit of openness I must confess I'm not a voracious reader as our fearless leader is. I'm a thinker, relying on perceptions and relationships I've cultivated. I use my imagination to envision a better world and work to make it happen. I fail miserably, but am hopeful.
Thanks Rachel and company, I love these conversations with you fellow searchers for truth. I'm glad we are sharing our ideas.
Many great thoughts on this topic. There are many attributes of humans that make us unpredictable and visionaries. I doubt if an AI could write a story like "Father Sergius" by Tolstoy. Computers may never be able to dream, although current research on dreaming as memory sorting may mimic such an experience.
It's even easier than the question: Being human individuals.
I would say that doubt is a part of the inductive reasoning process, which AI can learn to process. I reckon that the human skill that AI won't be able to emulate - at least not to the depth & extent that some humans have it - is 'inventiveness'.. I could be wrong though, but my understanding of AI is that it has to 'learn' from humans; and inventiveness is the one thing that one can only learn to a limited extent. It appears that we all have it to a greater or lesser degree. Consider the life of a farmer: he has to 'make a plan' at least 3 times a day on average, improvising and urgently solving new problems like a rusty irrigation valve breaks and he has to 'make a plan' real quick; a fire-break must be urgently burned at exactly the right strip of grass for the current wind direction; his borehole pump breaks and the local supplier quotes him 2 weeks delivery on a replacement and he has to 'do something' that will supply water for the next 2 weeks... it's that "make-a-plan" with a piece of wire and a used handkerchief and the lid of the wife's washing-basket factor that I don't believe AI will ever do as well as humans.. I may be wrong though
Subjective Life Experience and Human Perception.
AI can run routines to question its conclusions, but I think the doubt you are talking about varies depending on the human's experience. AI can mimic experience, but not fully incorporate a life long human experience. AI currently, has its limit, whereas human doubt changes over time, based on an individual's experience with others and the world. FAITH is my answer.
Doubt is a great one! I’ll add compassion.
Judgement. Humans are able to make ethical and moral decisions, in ways AI will never be able to do.
Truth. Large language model like ChatGPT predicts what words go together based on input text and all the material it has been trained on. It doesn't care if the information it serves up is true or factual in any sense of the word.
Interesting article my friend John Schneider found for me.
https://archive.vn/x9N7X