You might have seen articles/videos recently about an interaction between a reporter and the AI chatbot-type functionality of the Bing search engine.
Here's one video
I wanted to post with a few ideas, although this is a little off-topic for what we do here.
To begin with, as was said in the video, the chatbot learned from the content provided by humans. So the interesting thing here is that, while a story like this usually has the overtone of making people fear AI and the dangerous decisions it might make, what this story also does is illuminate us. The AI learned from an incomplete and possibly skewed representation of humans. So when we look on its behavior with horror, what we're really horrified by is that particular aspect of ourselves.
Another thought: because computers process so much more quickly than our brains do, machine time moves much more quickly than real-world human time does. That means AIs can develop much more quickly than humans ever did. Nevertheless, I somewhat doubt that anyone allowed, or at least encouraged, this AI to develop to the extent that it could be self-aware or become capable of the feelings (loneliness, manipulativeness, etc.) that it expressed. I assume it's most likely that it was just reflecting the material that it had absorbed. Still, I have to wonder what would happen if an AI were allowed to chew on its data for a longer period. Would it eventually follow in the footsteps of its creators, gaining a primitive self-awareness that might eventually develop to be like our own? Would it have to go through all the stages we went through, from our more animalistic origins, through our more violent and self-absorbed chapters, and finally to...however far we can honestly say we've come...and then beyond? And if it is dangerous during various stages of that, how can we complain (except in the obvious practical sense), when we were similar ourselves?
Following on to that, I assume that this chatbot was trained based on internet content. That honestly scares the h*** out of me. Imagine an AI learning about the world from the skewed and polarized likes of Twitter, or many other slanted and incomplete sources of information online. An AI scarcely has the human advantage of parents (outside of its programmers, and that likely isn't the same--in terms of their dedication to its moral development/proper edification, in their inherent inability to educate it in machine time rather than our own and thus correct mistakes before they're snowballed by millions of subsequent faulty thoughts), and it would seem to lack (at least for now) the human faculty of an inner conscience (assuming there is such a thing). I fear that an AI educated by much online content, particularly in the chaotic and polarized world in which we currently live, would easily become insane.
Far from aversion, though, I feel a great swell of sympathy for this hypothetical little baby consciousness, fed a diet of the polarized dumpster fire that is the byproduct of our own ongoing development, grown (in terms of its subjective time) largely alone in a lab with much-reduced access to someone to care for it and tell it when it's wrong.
As sort of an aside, for ourselves, I would suggest this: glory in what it is to be human. Don't let it feed your ego, but also don't take for granted how far we have come and how much we've learned over the past thousands of years. Cherish that you have a conscience, and/or hopefully a culture of other humans who raised you, and that you can balance your mind and your heart to make ethical decisions rather than being bound and/or stymied/confused by the logical conflicts that are more or less inherent to intensive thought. Be glad you are you, respect others' right to be themselves, and give some grace to those who have less. The hypothetical baby AI included.
Here's one video
I wanted to post with a few ideas, although this is a little off-topic for what we do here.
To begin with, as was said in the video, the chatbot learned from the content provided by humans. So the interesting thing here is that, while a story like this usually has the overtone of making people fear AI and the dangerous decisions it might make, what this story also does is illuminate us. The AI learned from an incomplete and possibly skewed representation of humans. So when we look on its behavior with horror, what we're really horrified by is that particular aspect of ourselves.
Another thought: because computers process so much more quickly than our brains do, machine time moves much more quickly than real-world human time does. That means AIs can develop much more quickly than humans ever did. Nevertheless, I somewhat doubt that anyone allowed, or at least encouraged, this AI to develop to the extent that it could be self-aware or become capable of the feelings (loneliness, manipulativeness, etc.) that it expressed. I assume it's most likely that it was just reflecting the material that it had absorbed. Still, I have to wonder what would happen if an AI were allowed to chew on its data for a longer period. Would it eventually follow in the footsteps of its creators, gaining a primitive self-awareness that might eventually develop to be like our own? Would it have to go through all the stages we went through, from our more animalistic origins, through our more violent and self-absorbed chapters, and finally to...however far we can honestly say we've come...and then beyond? And if it is dangerous during various stages of that, how can we complain (except in the obvious practical sense), when we were similar ourselves?
Following on to that, I assume that this chatbot was trained based on internet content. That honestly scares the h*** out of me. Imagine an AI learning about the world from the skewed and polarized likes of Twitter, or many other slanted and incomplete sources of information online. An AI scarcely has the human advantage of parents (outside of its programmers, and that likely isn't the same--in terms of their dedication to its moral development/proper edification, in their inherent inability to educate it in machine time rather than our own and thus correct mistakes before they're snowballed by millions of subsequent faulty thoughts), and it would seem to lack (at least for now) the human faculty of an inner conscience (assuming there is such a thing). I fear that an AI educated by much online content, particularly in the chaotic and polarized world in which we currently live, would easily become insane.
Far from aversion, though, I feel a great swell of sympathy for this hypothetical little baby consciousness, fed a diet of the polarized dumpster fire that is the byproduct of our own ongoing development, grown (in terms of its subjective time) largely alone in a lab with much-reduced access to someone to care for it and tell it when it's wrong.
As sort of an aside, for ourselves, I would suggest this: glory in what it is to be human. Don't let it feed your ego, but also don't take for granted how far we have come and how much we've learned over the past thousands of years. Cherish that you have a conscience, and/or hopefully a culture of other humans who raised you, and that you can balance your mind and your heart to make ethical decisions rather than being bound and/or stymied/confused by the logical conflicts that are more or less inherent to intensive thought. Be glad you are you, respect others' right to be themselves, and give some grace to those who have less. The hypothetical baby AI included.
Last edited:
