I think this mindset is part of the issue. You cannot converse with a deceased historical figure. AI does not change that. If you want to know what they said on a topic, you might be able
to search through their work, or read others interpretations of their work, but there is no way to ask them something they did not previously address and get an honest answer. You may be able to get an educated guess, but there is no way to know someone’s thoughts unless they were somehow already expressed. If you’re ok with AI just guessing as to what they might have said, then how are you to distinguish what they actually believed from things the AI just decided they believed. If you’d require the AI to add a disclaimer (“I never spoke on that, but if I had to guess…”) then who decides what follows? There’s plenty of debate on how to interpret certain works already, so which side does it tell? Does it do both? Is it going to both sides everything? I don’t think that would be a positive.
I understand wanting to be immersed in history, but I think original sources are our best way of doing so. I think movies or historical fiction books provide the necessary distance so there’s no real confusion. If you watch a dramatization of a historical figures life I don’t think people assume that every line is a direct quote. That kind of media also keeps the interaction on rails, so there’s no need to answer if George Washington would have enjoyed a brat summer. With AI people can often get whatever answer they want while using leading questions. If all of that was locked down, you’d be left with a fancy audiovisual quoting preexisting first party sources that just removes the impetus for the user to do any kind of sourcing themselves, while wasting time and energy.
A good 25 or so years ago I went to a one-man show on Mark Twain portrayed by Hal Holbrook. It was so great. He had hours of primary source material memorized. As a Lit Major who spent weeks digging through University libraries, and lifelong Mark Twain fan, it was one of the best things I’ve seen.
As I said in my original comment, I like the potential, but it’s not there yet. Anyone in this community knows the current state of hallucinating bullshit AI (Which is why I added firmly grounded and based on historical documents. I thought that was pretty clear, but I guess not). I was in no way endorsing that, and the projecting mindset that I was is a huge part of the problem with discussing just about anything these days.
EDIT: People don’t seem to understand what the word potential means.
I think their point is that it won’t ever be there/them. What makes a person can not be consolidated down into written documentation.
To add to that, AI is technically there. You can make an interactive image of someone (3D may not be there yet) with a custom voice and have it respond like the original person (based on info we have on them). The issue is that it will never actually be that person.
In 1903, the president of Michigan Savings Bank warned Henry Ford’s lawyer, Horace Rackham, to protect his money. “The horse is here to stay, but the automobile is only a novelty, a fad,” he advised.
The sum of a man is not a handful of primary sources. Unless we invent some form of time machine there isn’t a way to accurately portray historical figures. Anything that does will be influenced and biased by the person who created it (and will most likely be done to push a narrative)
And you seem to have misconstrued my point. We seem to be talking at odds without listening to each other, so it’s probably a good idea just to end it. I could just paraphrase your statement but substitute books (or whatever). It’s not the real thing, it’s going through a human filter that is going to have influenced and be biased, so it doesn’t matter. And things will never get good enough to prove me wrong. I’m done.
I think that kind of thing works best as an actor who has put in the work bringing that human element to it, like that Mark Twain thing you went to. It keeps that sense of distance, so you know it’s not 100% that historical figure (since I guarantee some people would absolutely believe that an ai “George Washington” is somehow like an actual scan of George Washington’s brain). I also think a human can bring that passion and understanding to the performance that an ai just can’t do, even if it’s an advanced, not-constantly-hallucinating ai
In 1904, the New York Times reported on a debate in Paris between a brain specialist and a physician about the dangers of driving automobiles at high speeds—because the brain can’t keep up. “It remains to be proved how fast the brain is capable of traveling,” reads the article. “If it cannot acquire an eight-mile per hour speed, then an auto running at the rate of 80 miles per hour is running without the guidance of the brain, and the many disastrous results are not to be marveled at.”
I have also enjoyed two similar shows, not mark twain, but also one man shows embodying a historical figure, apparently it’s a popular thing to do. I am glad to hear you had a positive experience like that.
I hope you did not feel I was taking your comment in bad faith. I believe I understood where you were coming from, but I think though the desire is valid, there’s no real way for AI to ever accomplish that. I understand that you want it to be grounded in fact, and I just think that in order to ground things in fact, it has to exclusively quote the original documents.
If I asked a perfectly modeled George Washington if the US should have been created, one could argue the AI Washington could reasonably say “yes”. If I then explained to that AI Washington that the country actually went on to create political factions exactly as he warned against and then provided a list of atrocities committed by the US and said they would not have been committed if we were under British rule, and then also explained how the British government has changed since he passed, and asked if he still supported the forming of the US, what is the AI going to say? One could make arguments multiple ways and even a “perfect” AI would never actually know. People can hold contradictory opinions or have their minds changed in unpredictable ways when given new information. My point is that historical documents are just that. Historical documents. They are not a perfect representation of the person and we should never expect them to be. To model an AI off of them and suggest talking to it is talking to the original person is in my opinion a falsehood and always will be no matter how good the AI is. I hear that you don’t think we’re there yet, but I would suggest that even if every moment of someone’s life could be analyzed by AI, there is no talking to them through the AI. Only a guess at what they might have said. That may be fine for some people, but I think it’s not positive to have people believe they definitely know what Washington would say about the marvel cinematic universe. I thought my interpretation was clear, and I’m now seeing a different comment that seems to have at least understood where I was coming from. They were much more concise lol.
If the AI is based in fact, and not allowed to hallucinate bullshit, than asking George “What do you think of the internet?” should get a response like “Sorry, out of bounds error”.
I think this mindset is part of the issue. You cannot converse with a deceased historical figure. AI does not change that. If you want to know what they said on a topic, you might be able to search through their work, or read others interpretations of their work, but there is no way to ask them something they did not previously address and get an honest answer. You may be able to get an educated guess, but there is no way to know someone’s thoughts unless they were somehow already expressed. If you’re ok with AI just guessing as to what they might have said, then how are you to distinguish what they actually believed from things the AI just decided they believed. If you’d require the AI to add a disclaimer (“I never spoke on that, but if I had to guess…”) then who decides what follows? There’s plenty of debate on how to interpret certain works already, so which side does it tell? Does it do both? Is it going to both sides everything? I don’t think that would be a positive.
I understand wanting to be immersed in history, but I think original sources are our best way of doing so. I think movies or historical fiction books provide the necessary distance so there’s no real confusion. If you watch a dramatization of a historical figures life I don’t think people assume that every line is a direct quote. That kind of media also keeps the interaction on rails, so there’s no need to answer if George Washington would have enjoyed a brat summer. With AI people can often get whatever answer they want while using leading questions. If all of that was locked down, you’d be left with a fancy audiovisual quoting preexisting first party sources that just removes the impetus for the user to do any kind of sourcing themselves, while wasting time and energy.
A good 25 or so years ago I went to a one-man show on Mark Twain portrayed by Hal Holbrook. It was so great. He had hours of primary source material memorized. As a Lit Major who spent weeks digging through University libraries, and lifelong Mark Twain fan, it was one of the best things I’ve seen.
As I said in my original comment, I like the potential, but it’s not there yet. Anyone in this community knows the current state of hallucinating bullshit AI (Which is why I added firmly grounded and based on historical documents. I thought that was pretty clear, but I guess not). I was in no way endorsing that, and the projecting mindset that I was is a huge part of the problem with discussing just about anything these days.
EDIT: People don’t seem to understand what the word potential means.
I think their point is that it won’t ever be there/them. What makes a person can not be consolidated down into written documentation.
To add to that, AI is technically there. You can make an interactive image of someone (3D may not be there yet) with a custom voice and have it respond like the original person (based on info we have on them). The issue is that it will never actually be that person.
You seem to have misconstrued my point.
The sum of a man is not a handful of primary sources. Unless we invent some form of time machine there isn’t a way to accurately portray historical figures. Anything that does will be influenced and biased by the person who created it (and will most likely be done to push a narrative)
And you seem to have misconstrued my point. We seem to be talking at odds without listening to each other, so it’s probably a good idea just to end it. I could just paraphrase your statement but substitute books (or whatever). It’s not the real thing, it’s going through a human filter that is going to have influenced and be biased, so it doesn’t matter. And things will never get good enough to prove me wrong. I’m done.
I think that kind of thing works best as an actor who has put in the work bringing that human element to it, like that Mark Twain thing you went to. It keeps that sense of distance, so you know it’s not 100% that historical figure (since I guarantee some people would absolutely believe that an ai “George Washington” is somehow like an actual scan of George Washington’s brain). I also think a human can bring that passion and understanding to the performance that an ai just can’t do, even if it’s an advanced, not-constantly-hallucinating ai
I have also enjoyed two similar shows, not mark twain, but also one man shows embodying a historical figure, apparently it’s a popular thing to do. I am glad to hear you had a positive experience like that.
I hope you did not feel I was taking your comment in bad faith. I believe I understood where you were coming from, but I think though the desire is valid, there’s no real way for AI to ever accomplish that. I understand that you want it to be grounded in fact, and I just think that in order to ground things in fact, it has to exclusively quote the original documents.
If I asked a perfectly modeled George Washington if the US should have been created, one could argue the AI Washington could reasonably say “yes”. If I then explained to that AI Washington that the country actually went on to create political factions exactly as he warned against and then provided a list of atrocities committed by the US and said they would not have been committed if we were under British rule, and then also explained how the British government has changed since he passed, and asked if he still supported the forming of the US, what is the AI going to say? One could make arguments multiple ways and even a “perfect” AI would never actually know. People can hold contradictory opinions or have their minds changed in unpredictable ways when given new information. My point is that historical documents are just that. Historical documents. They are not a perfect representation of the person and we should never expect them to be. To model an AI off of them and suggest talking to it is talking to the original person is in my opinion a falsehood and always will be no matter how good the AI is. I hear that you don’t think we’re there yet, but I would suggest that even if every moment of someone’s life could be analyzed by AI, there is no talking to them through the AI. Only a guess at what they might have said. That may be fine for some people, but I think it’s not positive to have people believe they definitely know what Washington would say about the marvel cinematic universe. I thought my interpretation was clear, and I’m now seeing a different comment that seems to have at least understood where I was coming from. They were much more concise lol.
If the AI is based in fact, and not allowed to hallucinate bullshit, than asking George “What do you think of the internet?” should get a response like “Sorry, out of bounds error”.