A friend once told me that top secret U.S. military technology is typically 10 to 20 years more advanced than the public and U.S. enemies are otherwise led to believe. I'd assume the same is true for AI. Particularly for written conversation. I also read, on a number of occasions, AI developers are highly cognizant that while they already can make AI human-like looking robots, they are typically asked to avoid making them too real because this might create a backlash. There is a State Farm insurance commercial that typifies this. A human like robot insurance agent is laughably made to look somewhat like the State Farm insurance agent, and when told it can't be as compassionate as a human, it argues it can, and proceeds to comically squirt tears out its eye sockets, and hugs the car while stating, "I'm so sorry for your au-to-mo-bile."
It's a funny commercial, but what some people fail/forget to realize is that it's not a robot. It's a human made up to look like a robot, and James Harden is a robot* (I'm kidding here,* but come on, how does he score 36 ppg in this league?), but because the technology is 20 years more advanced, we don't realize....
Anyway, I don't think we are too far off from the Turing test -- the hardest thing to replicate might be touch. I'm certain an AI could make it through more than twenty questions via written form. If we imagine voice inflection is ten years more advanced than we know, I'm pretty certain they could fool us over the phone for a good two to three minutes as well. In person? Well, the current dancing, walking, singing robots definitely look more artificial than Tammy Faye Bakker, but just consider how much more real Madam Tussad's replicas are becoming, then imagine another ten to twenty years into the future. They will get better and better and better. Kurzweil once said our expectations overshoot what we think technology can achieve in a five year period, but underestimate what actually occurs in a ten year period. Ray is responsible (partly/mostly) for those voice identification programs that can transcribe in real time.
On this board, if there are any AI, it typically uses two primary covering methods -- anger and short answers. Although extremely long rambling answers seem par for the course as well. LOL.
Respect Risk. Conduct Your Own Due Diligence. Manage your assets wisely. Diversify.
