Excellent essay. You seem to have hit on the Archimedean point needed for looking at this thing. But, I balked when you stated we entered into a "I-It" relationship with these chatbots. It seems to me it is just the opposite, and the problem is that for all its "I-You" appearances, the chatbot is an "it". Furthermore, it is an "it" where we don't control it, and we worry its producers might not either. But, that feels like a red herring to me. If mass statistics based on what we do online are much more enriching and controlling than statistics based on what we tell the sampler we will do, just think how much more enriching and controlling still will be statistics based on what we are actually thinking. That seems like a problem that is here and now.
Hi Tom, thanks for commenting. I think we agree about chatbots representing an I-It relationship. And you’re right, I haven’t been very clear about what the human relationship with a genuinely autonomous AGI might mean. I think I’m still ambivalent about that possibility. My main concern is that, as AI demonstrates greater and greater autonomy from human decisions, we will be too quick to persuade ourselves that we have discovered a being with which we can have the kind of meaningful and responsive relationship that is otherwise difficult to find under conditions of modernity. I am inclined to think that, rather than discovering a new object of genuine relation, we are much more likely, in our desperation, to be misled by a technology that remains essentially a reflection of ourselves—and believing ourselves to have engineered a solution to modern world-alienation, we may be less inclined to look to older sources of authentic encounters with the world.
Excellent essay. You seem to have hit on the Archimedean point needed for looking at this thing. But, I balked when you stated we entered into a "I-It" relationship with these chatbots. It seems to me it is just the opposite, and the problem is that for all its "I-You" appearances, the chatbot is an "it". Furthermore, it is an "it" where we don't control it, and we worry its producers might not either. But, that feels like a red herring to me. If mass statistics based on what we do online are much more enriching and controlling than statistics based on what we tell the sampler we will do, just think how much more enriching and controlling still will be statistics based on what we are actually thinking. That seems like a problem that is here and now.
Hi Tom, thanks for commenting. I think we agree about chatbots representing an I-It relationship. And you’re right, I haven’t been very clear about what the human relationship with a genuinely autonomous AGI might mean. I think I’m still ambivalent about that possibility. My main concern is that, as AI demonstrates greater and greater autonomy from human decisions, we will be too quick to persuade ourselves that we have discovered a being with which we can have the kind of meaningful and responsive relationship that is otherwise difficult to find under conditions of modernity. I am inclined to think that, rather than discovering a new object of genuine relation, we are much more likely, in our desperation, to be misled by a technology that remains essentially a reflection of ourselves—and believing ourselves to have engineered a solution to modern world-alienation, we may be less inclined to look to older sources of authentic encounters with the world.