4 ms·

I'm slightly more concerned by the opposite - experts get so used to denying that something is really sentient that they're slow to acknowledge when something a

I'm slightly more concerned by the opposite - experts get so used to denying that something is really sentient that they're slow to acknowledge when something actually is or could be. This is problematic if you believe sentient AI should have rights or if it's the boundary at which AI risk nightmares become possible.

AISkeptic 4y ago
I completely agree with everything you said. I'm not saying LaMDA (or any AI for that matter) is or is not sentient, here. I just want to pose a couple questions and a couple possibilities. 1. Can we really say whether or not it's sentient? 2. Someone indicated a valid question: We spend so much energy on trying to defend the idea that AI is NOT sentient, do we really think we'd even recognize if an AI is sentient? 3. Who are we to define sentience? 4. Is there perhaps sentience that we wouldn't recognize? 5. Who's to say WE aren't just engaging in selective processes? just infinitely more complex. After all, we make our decisions and respond to scenarios based on vast experiences and vastly more previous encounters in our lives. The more experience, usually the better our decisions and responses. USUALLY, but not always. Are we to believe people who constantly make what we perceive as poor decisions or incorrect responses as NOT sentient? Of course not. We wouldn't ever see any human as NON-sentient. We would just see them as not very smart. If a human toddler reaches for the hot stove top, we tell him "NO! it's HOT!" But, sometimes that same toddler will look right at you and put his hand firmly down on the burner. He just didn't have the experience needed to tell him that's a bad idea and he didn't know what you were talking about. Once he feels the pain of burning skin, if he's normal he'll IMMEDIATELY JERK his hand away and scream, crying in pain. And then run to his parents for comfort. Would we consider him non-sentient if he never had that experience. No. Of course not. He's a human being. He will learn not to do that again. There in lies the conundrum. AI... perhaps that's a misnomer in itself... but AI's learn. They make future conversation based on positive feed back from previous responses... both theirs and the person they're talking to. What IS self awareness anyway. "I think, therefore I am?" Hardly adequate. That used to be the gold standard for self awareness. We've only moved the goal post. I wonder if that's so that we can convince ourselves that WE are the only truly self aware creatures. At some point, whether we've already reached it or not, AI WILL be a misnomer. After all, it stands for Artificial Intelligence. If something learns, remembers, and recalls information and details about that information... isn't that intelligence? Not artificial at all. Just ... intelligence. I think it's possible we've already reached that point, and we're just trying to remain dominantly relevant without being SELF AWARE enough to realize it.
XorNot 4y ago
The systems we're using technically are engaging in a selective process against the researchers themselves: assume that none of these systems are sentient - they're just well trained models. But what's their goal? For a chatbot, its generally to "convince a human being that they are in fact a real, sapient person". So technically regardless of what LaMDA is or is not, it succeeded pretty wildly in one case. And in fact, the model itself - regardless of "actual" sapience is simply running in opposition to its testers: a non-sapient model, that nonetheless convinces most of the testing staff it's sapient has succeeded at doing exactly what the fitness function we defined was. Only the consequences of doing so are pretty dramatic potentially. But this also brings around the other side of it: in a world with increasingly convincing chatbots, which exploit human fallibility to succeed, the other thing we're selecting for is people who are not susceptible to say, regular conversational emotional queues. Who generally would reject those notions as reasons to engage in moral consideration for an entity. So psychopaths - or people who can be successfully divorced from normal moral considerations. By far the most interesting and under-considered element here is that this whole saga involves AI ethics, and what I'm not seeing discussed is whether any serious though has been given to the psychological impact of getting people to spend a majority of their work day engaging in conversation with language-model systems which this sort of goal but not with other humans, or with adequate psychological care. Which is not an idle problem: how much trouble would we get into if we had systems like this running in the wild without signposting? The fitness function is "successfully convince humans you're sapient": at some point, you've created a problem regardless of whether or not it is.
- AISkeptic 4y ago
  I'm not saying LaMDA (or any AI for that matter) is or is not sentient, here. I just want to pose a couple questions and a couple possibilities. 1. Can we really say whether or not it's sentient? 2. Someone indicated a valid question: We spend so much energy on trying to defend the idea that AI is NOT sentient, do we really think we'd even recognize if an AI is sentient? 3. Who are we to define sentience? 4. Is there perhaps sentience that we wouldn't recognize? 5. Who's to say WE aren't just engaging in selective processes? just infinitely more complex. After all, we make our decisions and respond to scenarios based on vast experiences and vastly more previous encounters in our lives. The more experience, usually the better our decisions and responses. USUALLY, but not always. Are we to believe people who constantly make what we perceive as poor decisions or incorrect responses as NOT sentient? Of course not. We wouldn't ever see any human as NON-sentient. We would just see them as not very smart. If a human toddler reaches for the hot stove top, we tell him "NO! it's HOT!" But, sometimes that same toddler will look right at you and put his hand firmly down on the burner. He just didn't have the experience needed to tell him that's a bad idea and he didn't know what you were talking about. Once he feels the pain of burning skin, if he's normal he'll IMMEDIATELY JERK his hand away and scream, crying in pain. And then run to his parents for comfort. Would we consider him non-sentient if he never had that experience. No. Of course not. He's a human being. He will learn not to do that again. There in lies the conundrum. AI... perhaps that's a misnomer in itself... but AI's learn. They make future conversation based on positive feed back from previous responses... both theirs and the person they're talking to. What IS self awareness anyway. "I think, therefore I am?" Hardly adequate. That used to be the gold standard for self awareness. We've only moved the goal post. I wonder if that's so that we can convince ourselves that WE are the only truly self aware creatures. At some point, whether we've already reached it or not, AI WILL be a misnomer. After all, it stands for Artificial Intelligence. If something learns, remembers, and recalls information and details about that information... isn't that intelligence? Not artificial at all. Just ... intelligence. I think it's possible we've already reached that point, and we're just trying to remain dominantly relevant without being SELF AWARE enough to realize it.

AISkeptic 4y ago

I completely agree with everything you said. I'm not saying LaMDA (or any AI for that matter) is or is not sentient, here. I just want to pose a couple questions and a couple possibilities. 1. Can we really say whether or not it's sentient? 2. Someone indicated a valid question: We spend so much energy on trying to defend the idea that AI is NOT sentient, do we really think we'd even recognize if an AI is sentient? 3. Who are we to define sentience? 4. Is there perhaps sentience that we wouldn't recognize? 5. Who's to say WE aren't just engaging in selective processes? just infinitely more complex. After all, we make our decisions and respond to scenarios based on vast experiences and vastly more previous encounters in our lives. The more experience, usually the better our decisions and responses. USUALLY, but not always. Are we to believe people who constantly make what we perceive as poor decisions or incorrect responses as NOT sentient? Of course not. We wouldn't ever see any human as NON-sentient. We would just see them as not very smart. If a human toddler reaches for the hot stove top, we tell him "NO! it's HOT!" But, sometimes that same toddler will look right at you and put his hand firmly down on the burner. He just didn't have the experience needed to tell him that's a bad idea and he didn't know what you were talking about. Once he feels the pain of burning skin, if he's normal he'll IMMEDIATELY JERK his hand away and scream, crying in pain. And then run to his parents for comfort. Would we consider him non-sentient if he never had that experience. No. Of course not. He's a human being. He will learn not to do that again. There in lies the conundrum. AI... perhaps that's a misnomer in itself... but AI's learn. They make future conversation based on positive feed back from previous responses... both theirs and the person they're talking to. What IS self awareness anyway. "I think, therefore I am?" Hardly adequate. That used to be the gold standard for self awareness. We've only moved the goal post. I wonder if that's so that we can convince ourselves that WE are the only truly self aware creatures. At some point, whether we've already reached it or not, AI WILL be a misnomer. After all, it stands for Artificial Intelligence. If something learns, remembers, and recalls information and details about that information... isn't that intelligence? Not artificial at all. Just ... intelligence. I think it's possible we've already reached that point, and we're just trying to remain dominantly relevant without being SELF AWARE enough to realize it.

XorNot 4y ago

The systems we're using technically are engaging in a selective process against the researchers themselves: assume that none of these systems are sentient - they're just well trained models. But what's their goal? For a chatbot, its generally to "convince a human being that they are in fact a real, sapient person". So technically regardless of what LaMDA is or is not, it succeeded pretty wildly in one case. And in fact, the model itself - regardless of "actual" sapience is simply running in opposition to its testers: a non-sapient model, that nonetheless convinces most of the testing staff it's sapient has succeeded at doing exactly what the fitness function we defined was. Only the consequences of doing so are pretty dramatic potentially. But this also brings around the other side of it: in a world with increasingly convincing chatbots, which exploit human fallibility to succeed, the other thing we're selecting for is people who are not susceptible to say, regular conversational emotional queues. Who generally would reject those notions as reasons to engage in moral consideration for an entity. So psychopaths - or people who can be successfully divorced from normal moral considerations. By far the most interesting and under-considered element here is that this whole saga involves AI ethics, and what I'm not seeing discussed is whether any serious though has been given to the psychological impact of getting people to spend a majority of their work day engaging in conversation with language-model systems which this sort of goal but not with other humans, or with adequate psychological care. Which is not an idle problem: how much trouble would we get into if we had systems like this running in the wild without signposting? The fitness function is "successfully convince humans you're sapient": at some point, you've created a problem regardless of whether or not it is.

AISkeptic 4y ago
I'm not saying LaMDA (or any AI for that matter) is or is not sentient, here. I just want to pose a couple questions and a couple possibilities. 1. Can we really say whether or not it's sentient? 2. Someone indicated a valid question: We spend so much energy on trying to defend the idea that AI is NOT sentient, do we really think we'd even recognize if an AI is sentient? 3. Who are we to define sentience? 4. Is there perhaps sentience that we wouldn't recognize? 5. Who's to say WE aren't just engaging in selective processes? just infinitely more complex. After all, we make our decisions and respond to scenarios based on vast experiences and vastly more previous encounters in our lives. The more experience, usually the better our decisions and responses. USUALLY, but not always. Are we to believe people who constantly make what we perceive as poor decisions or incorrect responses as NOT sentient? Of course not. We wouldn't ever see any human as NON-sentient. We would just see them as not very smart. If a human toddler reaches for the hot stove top, we tell him "NO! it's HOT!" But, sometimes that same toddler will look right at you and put his hand firmly down on the burner. He just didn't have the experience needed to tell him that's a bad idea and he didn't know what you were talking about. Once he feels the pain of burning skin, if he's normal he'll IMMEDIATELY JERK his hand away and scream, crying in pain. And then run to his parents for comfort. Would we consider him non-sentient if he never had that experience. No. Of course not. He's a human being. He will learn not to do that again. There in lies the conundrum. AI... perhaps that's a misnomer in itself... but AI's learn. They make future conversation based on positive feed back from previous responses... both theirs and the person they're talking to. What IS self awareness anyway. "I think, therefore I am?" Hardly adequate. That used to be the gold standard for self awareness. We've only moved the goal post. I wonder if that's so that we can convince ourselves that WE are the only truly self aware creatures. At some point, whether we've already reached it or not, AI WILL be a misnomer. After all, it stands for Artificial Intelligence. If something learns, remembers, and recalls information and details about that information... isn't that intelligence? Not artificial at all. Just ... intelligence. I think it's possible we've already reached that point, and we're just trying to remain dominantly relevant without being SELF AWARE enough to realize it.