It was, at the least, an unnecessary medical procedure that resulted in someone's death. So you arrest him for homicide, seize the equipment and reveal the fact that it is just some kind of electric chair. It'll go to trial and the lies will be revealed, or he'll have to give a complete confession.
Otherwise I think this is a bit of a trick question. We know that we'd be able to tell the difference between the deepfake and a specific living person -- they have a lifetime of memories, a multitude of unique relationships with various people, the ability to effortlessly distinguish a chihuahua from a blueberry muffin, etc. But if the premise is that the salesman's simulation is so good that we can't tell the difference at all, then is there actually any lie to detect?
> So you arrest him for homicide, seize the equipment and reveal the fact that it is just some kind of electric chair. It'll go to trial and the lies will be revealed, or he'll have to give a complete confession.
There may be environments today in which prosecution under the current regime would not be possible. What if he was doing this in a hospital, and had full legal immunity granted similar to Pfizer?
> We know that we'd be able to tell the difference between the deepfake and a specific living person
After seeing the advanced deep learning systems OpenAI has put out there, like GPT-3 and DALL-E, I'm not so sure. No, those systems aren't perfect, but the imperfections could be hidden from view in a low-fidelity virtual environment like FB's metaverse.
> But if the premise is that the salesman's simulation is so good that we can't tell the difference at all, then is there actually any lie to detect?
For myself and many people, human lives have a special status, and they more than just an algorithm or input/output relationship.
The idea that a simulation of a person, no matter how well-trained, being morally or legally equivalent to that person, just does not compute to people like me.
Also, a liar knows full well that he is lying -- so why should other people be confused about it, once they know it?
Otherwise I think this is a bit of a trick question. We know that we'd be able to tell the difference between the deepfake and a specific living person -- they have a lifetime of memories, a multitude of unique relationships with various people, the ability to effortlessly distinguish a chihuahua from a blueberry muffin, etc. But if the premise is that the salesman's simulation is so good that we can't tell the difference at all, then is there actually any lie to detect?