Xiaoice listens. That is, if we're Chinese, when we send her a text message, she responds in a way that seems like listening. Pronounced shao-ice, she is the persona of a Microsoft chatbot with a reported 20 million registered users.
Remembering details of previous conversation sessions and making use of words and phrases mined from the wealth of Chinese internet conversations, Xiaoice, to all reports, seems up-to-date, attentive, and often fun. Lots of young Chinese look forward to the 'voice' associated with the program that some refer to as their 'life partner.'
Does conversation with a chatbot qualify as a 2nd person encounter? Author John Markoff remarks in his article in the New York Times, 'Such programs represent meaning as elaborate statistical relationships between words, sentences and objects. These equations are refined as millions of images or utterances are added to the database, improving the programs' ability to accurately recognize patterns.' If meaning is so defined, can such a program have thought projects (as opposed to being a thought project)? Is it able to question premises, can it recognize otherness, does it create new forms or genres? Is there a second person at all?
Yet we do address all kinds of things as 'you', attributing a kind of 2nd person-ness to them, and in the process learn something about these objects through our scrutiny, and something about ourselves through our eavesdropping on ourselves. Yet we are aware of the difference between encounters of reaction, of response and of reply.
A reaction is a simple consequence of an action, e.g. the shudder of a building when a car runs into it. A response is a consequence expressive of intentionality, e.g. the movement of the wheels when I turn the steering wheel. The root of the word is 'spondere' meaning 'to pledge' which I think refers to the idea of a dedicated outcome. A reply is a response that addresses the originator, e.g. an answer to a question. The root of the word is 'plicare' meaning 'to fold' referring, I think, through the concept of 'braid' to that of implicit intertwined existence in a context of mutual acknowledgement.
So when we address an oyster, a tomato, a boat, or Lady Justice or some such, we're not really disappointed when we don't get a reply, other than the Other being itself, and reacting or responding as it would normally or was designed to do. If, however, something on the same order is not just behaving, but purportedly replying, it's making a claim about implication that can't be called insincere (unless in reference to the intentions of the designers) but is false. Xiaoice is not invested in its text partners as they in it, in fact, is not invested at all. They weave her into their lives; she does not, can not weave them into hers. She is a presence that is not a presence.
So what? Well, an encounter with a chatbot may be in the mode of reaction or response, but not, I think, of reply for which individuals of independent point of view, free to assent, refuse or propose, and capable of explicit mutuality, are required if an encounter is to be really rich.
And beyond that, the passionate engagement of God-in-love and the beloved Other in which we participate can serve as a model of the heights and depths our encounters can aspire to: dances of hospitality, friendship and exploration, with 'each longing for and delighting in the presence of the other; each interested in what interests the other, what the other chooses to do--and how; each loyal to the fulfillment of the other; each willing to take the place of the other in pain and to give place to the other in joy and their passionate conversation--the whole universe of ever-evolving complexity drawn in, shared and appreciated within it...'
An interaction with Xiaoice may be an image of such an encounter, and images have some value; Microsoft is looking for ways of commercializing the program.
Remembering details of previous conversation sessions and making use of words and phrases mined from the wealth of Chinese internet conversations, Xiaoice, to all reports, seems up-to-date, attentive, and often fun. Lots of young Chinese look forward to the 'voice' associated with the program that some refer to as their 'life partner.'
Does conversation with a chatbot qualify as a 2nd person encounter? Author John Markoff remarks in his article in the New York Times, 'Such programs represent meaning as elaborate statistical relationships between words, sentences and objects. These equations are refined as millions of images or utterances are added to the database, improving the programs' ability to accurately recognize patterns.' If meaning is so defined, can such a program have thought projects (as opposed to being a thought project)? Is it able to question premises, can it recognize otherness, does it create new forms or genres? Is there a second person at all?
Yet we do address all kinds of things as 'you', attributing a kind of 2nd person-ness to them, and in the process learn something about these objects through our scrutiny, and something about ourselves through our eavesdropping on ourselves. Yet we are aware of the difference between encounters of reaction, of response and of reply.
A reaction is a simple consequence of an action, e.g. the shudder of a building when a car runs into it. A response is a consequence expressive of intentionality, e.g. the movement of the wheels when I turn the steering wheel. The root of the word is 'spondere' meaning 'to pledge' which I think refers to the idea of a dedicated outcome. A reply is a response that addresses the originator, e.g. an answer to a question. The root of the word is 'plicare' meaning 'to fold' referring, I think, through the concept of 'braid' to that of implicit intertwined existence in a context of mutual acknowledgement.
So when we address an oyster, a tomato, a boat, or Lady Justice or some such, we're not really disappointed when we don't get a reply, other than the Other being itself, and reacting or responding as it would normally or was designed to do. If, however, something on the same order is not just behaving, but purportedly replying, it's making a claim about implication that can't be called insincere (unless in reference to the intentions of the designers) but is false. Xiaoice is not invested in its text partners as they in it, in fact, is not invested at all. They weave her into their lives; she does not, can not weave them into hers. She is a presence that is not a presence.
So what? Well, an encounter with a chatbot may be in the mode of reaction or response, but not, I think, of reply for which individuals of independent point of view, free to assent, refuse or propose, and capable of explicit mutuality, are required if an encounter is to be really rich.
And beyond that, the passionate engagement of God-in-love and the beloved Other in which we participate can serve as a model of the heights and depths our encounters can aspire to: dances of hospitality, friendship and exploration, with 'each longing for and delighting in the presence of the other; each interested in what interests the other, what the other chooses to do--and how; each loyal to the fulfillment of the other; each willing to take the place of the other in pain and to give place to the other in joy and their passionate conversation--the whole universe of ever-evolving complexity drawn in, shared and appreciated within it...'
An interaction with Xiaoice may be an image of such an encounter, and images have some value; Microsoft is looking for ways of commercializing the program.
No comments:
Post a Comment