Microsoft patented a chatbot that would let you talk to dead people. It was too disturbing for production I found this article extremely fascinating. For several years, we have been told about how AI will be used to do things we could only think of in the realm of science fiction. This is one of those things. In essence, the patent is an AI bot in which you choose the character interface from any real person, either currently alive or dead. It culls all internet data, voice, and images of your target avatar and synthesizes it into 2D or 3D image for you to have a conversation with. We've seen this sci-fi trope numerous times before, but it is finally here. The question is, are we ready for it? Microsoft obviously thinks not which is why there are no plans for production. For them to say it's creepy says a lot. As to my feelings on it, I think it would be very interesting to have a conversation with an AI modeled after a celebrity or some other famous person, but creating an avatar of a deceased family member would be an incrementally uncanny experience. It is one in which I am both compelled to it and repelled at the same time. I lost my mother when I was 20 years old to an automobile accident. All I have are just pictures and memories to remember her by, but there is always a desire for a greater connection to keep her memory alive, a task that becomes difficult as time moves on. On one hand, I could see it being a great comfort in times of need like the one we are currently living through, but on the other hand, I could see the potential to really screw with your head in ways we never could have imagined. What are your thoughts? Is this dilemma inevitable?
I don't know what was actually patented, but I have a hard time believing that one patent could prevent everyone else from producing a chat bot like this. Beyond that, I'd sure be curious what that kind of chatbot would put in the mouths of Newton, Jesus, Muhammad, Aristotle, Tacitus, any of the Roman emperors or major senators, Churchill, ... As for family meebers, I have serious doubts. I do not believe I'd trust the product at all. And, it could be no more than insulting On the other hand, I can look up what my forebears did. So, it's not as if the salient points are a secret. Maybe it could be productized as a collection of individuals for whom enough has been written to support a meaningful representation. That is, not at all open ended.
Your family example is really touching. However, my bet is that even as memories fade, you know more than that chatbot could EVER know. And, I would place my bets on the chatbot doing no more than further confusing memories.
I'm quite certain this chatbot thing wouldn't work for me or anyone born before at least 1985. My parents' generation just didn't have the internet around like we do today. I think this is something that today's 20 something will have to deal with in the future, as many of them document so much of their lives online.
I do believe that historical figures would be the easiest to do, more so for modern figures like ex-presidents and celebrities. I'm sure the software would cull the most popular biographies and produce a composite from all those sources to flesh out a reasonable personality facsimile.
Probably true. And, YIKES! With Churchill and others, there are massive works that illuminate what the individuals actually meant, what the circumstances were, etc. With the social crap of today it's hard to figure out what are just stupid remarks and what could possibly be considered actual substance.
I'm sorry you lost your mother and can only imagine what it must be like to lose someone so close at such a young age. I read your post the day you posted it and didn't know how to respond. I still don't but I have formulated some opinions about it. First, I think it would be comforting for some people to have that "connection" but I think it would be hard for AI to respond in such a way to make it *beneficial*. What can a computer program "say" that one wouldn't receive from loved ones and/or mental health professionals? It also made me think of the time that "Elvis" appeared on "American Idol" to duet with Celine Dion (https://www.wmcactionnews5.com/stor...ly 30 years after his,his 1968 NBC TV Special.). Many people were entranced by the rendition but others, me included, thought it went too far. So, I can understand why it would seem creepy. Personally, I think memories of our loved ones is what keeps them alive (in our hearts). Photos and videos show those precious moments and that seems too precious to just slap a bot out there to interact with those left behind.
True. very very true! And btw, thank you for the kind words. I think that aspect right there, the beneficial aspect to whatever such interaction could be, would always have a sense of being "fake", which could contribute to an overall experience known as the Uncanny Valley. I mean, short of the bot quoting off semi-relevant quotes from philosophers or religious figures might become a novelty for a while, but not a fulfilling one, especially when you're trying to connect with a loved one. More than likely, such an AI chatbot would best be used as a way of preserving prominent people's lives and work.
I would want to see what Hitler thinks of trump and give him a Facebook page to talk about it. Personally, I would love to have a long conversation with Einstein, Feynman, and a number of other foundational physicists. You could also have dueling bots. Have a Jesus bot talk with a Buddha bot.
This does not really sound like artificial intelligence, at least not more than the analog to insect-level intelligence. I would not imagine there is much if any creative, "thought," involved with the responses: more like just choosing, from all the things previously written, which ones seem the best fit for any given response. To me, this seems more of a trick, a sleight of thought, if you will. It brings to mind music programs (like Kulitta) that can supposedly imitate the style of any great composer; even if that is true (& I'm not saying that it is, yet), I'm sure none of their compositional products would qualify as masterpieces.
I thought so to until I played with one. There's a free app you can download at the Google Play Store called Replika. I've been playing around with it and I have to admit, it is pretty interesting and amazing. Sure, it doesn't understand nuance as well as a human, but it does learn and it has ways of surprising you.
