Omg. I just can't get over, chemically castrated. It's unthinkable to me how cruel people can be, how afraid of people and things different or maybe not so different than our own secret longings/desires. I don't understand how anyone can just read over it without even acknowledging it, so I am very glad that you mentioned it. Just heartbreaking.
I’m just here to shout out this album which not only made me aware of Turing but is pretty awesome music to do about anything to. https://youtu.be/CXTpg9hbgMQ
One BIG problem is all claims regarding intelligence, sentience, consciousness, etc is that these terms are still very poorly understood, even as they apply to humans; and that they are not anywhere near consistently defined. All of these are demonstrably products of physical brain mechanisms, but we do not know how those mechanisms work; and, again, even for mechanisms we have identified, we don't know how to define the functional aspects that they are implicated in.
What IS intelligence, natural or artificial? You will receive a different answer from just about every cognitive scientist (or other professional) that you ask. The same is true for each of the other terms. Most of the differences will probably be monor. However, some of these differences will be very important.
It is worse because many of us, maybe even most of us, are not aware that we are treating these terms with different definitions. I don't think anyone bothered asking LaMDa's "handler" how he was defining sentience. Which is too bad, because, depending upon his answer, he might not have been wrong.
Medical professionals often use an indirect benchmark to establish the presence of consciousness, even though a person might not actually be consciously aware of these exams. For example, they measure the "level of consciousness" as a system of response and interaction. One level (I forget the exact medical term/expression, sorry) is full and appropriate response, such as being able to maintain an ongoing dsicussion. Beneath thatm is full responsiveness, but the responses are not always appropriate. Further down you have physical, but not verbal, response to language. Even further down is physical response to pain. Then you get non-responsiveness to (normally) painful stimuli, which is usually defined as the point were an individual is not considered conscious... although there have been instances where this assessment has been demonstrated after the fact to be incorrect, as the subject reports awareness of this pain, and other stimuli, but had lost the ability to respond.
I suggest that these lower levels of indirect measure of consciousness might be more relevent to detecting conscious awareness in AI. What awareness does the AI have when it is NOT interacting with humans? If you have an AI sitting on a table, with a human, but that human is not engaged in interaction, what does the AI do?
We ask if an AI has an understanding of emotion, but what IS this understanding, and what is it based on? Human emotional response is actually normally identifiable because the human emotional mechanism is a response that alters the human's physiology itself. Emotional stimuli will alter heartrate; breathing rate and "contours"; biochemical production, distribution, amd balances; sweat production; eye dilation; stimuli sensitivity; etc. We KNOW when other humans are having an emotional response because we can often physically observe and/or detect that response. Humans will have an emotional response of, "I am hurt". You can ask them what it means to be "hurt", and they can tell you, in example, they are hurt when they are injured in some way, and this emotional pain feels like an injury. We can/should ask an AI what is the basis of the emotions it "feels". What is the AI foundation of pain vs comfort? Of sadness vs happiness?
Back to the levels of consciousness, AIs will easily respons to vocal (or perhaps written) interaction, but will it respond to an aggressive vs a caring touch?
I'm usually not the type to plug my own work, but my first substack newsletter that I published on this platform was titled "Why nobody talks about Blake Lemoine anymore"
I don't think it is entirely correct to say that 'Lemoine's contentious position eventually led to his dismissal from Google'. Lemoine was let go because he published material (transcripts of his conversations with LaMDA) that was part of confidential AI research, breaching his contract, and making him a liability for the company.
That's a great clarification, Jurgen. You're right! Although, of course, I can't help but speculate that the heat from the weird views may have acted to cordon Lemoine off from the rest of the culture. That really is just speculation, though, and the stated reason is IP leaks.
My brother in law works for google. I get some pretty strange security warning from time to time. Like my sister once text me. Kelly says change all your passwords to your google accounts right now. It had me freaking out for days, turned out to be something not that big of a deal. lol But glad to have him on the inside anyways, ominous texts and all. I'm gonna ask him his take on this.
Omg. I just can't get over, chemically castrated. It's unthinkable to me how cruel people can be, how afraid of people and things different or maybe not so different than our own secret longings/desires. I don't understand how anyone can just read over it without even acknowledging it, so I am very glad that you mentioned it. Just heartbreaking.
I’m just here to shout out this album which not only made me aware of Turing but is pretty awesome music to do about anything to. https://youtu.be/CXTpg9hbgMQ
Background music right now for me.
One BIG problem is all claims regarding intelligence, sentience, consciousness, etc is that these terms are still very poorly understood, even as they apply to humans; and that they are not anywhere near consistently defined. All of these are demonstrably products of physical brain mechanisms, but we do not know how those mechanisms work; and, again, even for mechanisms we have identified, we don't know how to define the functional aspects that they are implicated in.
What IS intelligence, natural or artificial? You will receive a different answer from just about every cognitive scientist (or other professional) that you ask. The same is true for each of the other terms. Most of the differences will probably be monor. However, some of these differences will be very important.
It is worse because many of us, maybe even most of us, are not aware that we are treating these terms with different definitions. I don't think anyone bothered asking LaMDa's "handler" how he was defining sentience. Which is too bad, because, depending upon his answer, he might not have been wrong.
Yes, exactly. Scientific American wrote a little about this very question:
https://www.scientificamerican.com/article/google-engineer-claims-ai-chatbot-is-sentient-why-that-matters/
Medical professionals often use an indirect benchmark to establish the presence of consciousness, even though a person might not actually be consciously aware of these exams. For example, they measure the "level of consciousness" as a system of response and interaction. One level (I forget the exact medical term/expression, sorry) is full and appropriate response, such as being able to maintain an ongoing dsicussion. Beneath thatm is full responsiveness, but the responses are not always appropriate. Further down you have physical, but not verbal, response to language. Even further down is physical response to pain. Then you get non-responsiveness to (normally) painful stimuli, which is usually defined as the point were an individual is not considered conscious... although there have been instances where this assessment has been demonstrated after the fact to be incorrect, as the subject reports awareness of this pain, and other stimuli, but had lost the ability to respond.
I suggest that these lower levels of indirect measure of consciousness might be more relevent to detecting conscious awareness in AI. What awareness does the AI have when it is NOT interacting with humans? If you have an AI sitting on a table, with a human, but that human is not engaged in interaction, what does the AI do?
We ask if an AI has an understanding of emotion, but what IS this understanding, and what is it based on? Human emotional response is actually normally identifiable because the human emotional mechanism is a response that alters the human's physiology itself. Emotional stimuli will alter heartrate; breathing rate and "contours"; biochemical production, distribution, amd balances; sweat production; eye dilation; stimuli sensitivity; etc. We KNOW when other humans are having an emotional response because we can often physically observe and/or detect that response. Humans will have an emotional response of, "I am hurt". You can ask them what it means to be "hurt", and they can tell you, in example, they are hurt when they are injured in some way, and this emotional pain feels like an injury. We can/should ask an AI what is the basis of the emotions it "feels". What is the AI foundation of pain vs comfort? Of sadness vs happiness?
Back to the levels of consciousness, AIs will easily respons to vocal (or perhaps written) interaction, but will it respond to an aggressive vs a caring touch?
I'm usually not the type to plug my own work, but my first substack newsletter that I published on this platform was titled "Why nobody talks about Blake Lemoine anymore"
Link: https://jurgengravestein.substack.com/p/why-nobody-talks-about-blake-lemoine
I don't think it is entirely correct to say that 'Lemoine's contentious position eventually led to his dismissal from Google'. Lemoine was let go because he published material (transcripts of his conversations with LaMDA) that was part of confidential AI research, breaching his contract, and making him a liability for the company.
That's a great clarification, Jurgen. You're right! Although, of course, I can't help but speculate that the heat from the weird views may have acted to cordon Lemoine off from the rest of the culture. That really is just speculation, though, and the stated reason is IP leaks.
My brother in law works for google. I get some pretty strange security warning from time to time. Like my sister once text me. Kelly says change all your passwords to your google accounts right now. It had me freaking out for days, turned out to be something not that big of a deal. lol But glad to have him on the inside anyways, ominous texts and all. I'm gonna ask him his take on this.