1. When we say we are trying to create an AI future, are we replicating one human or are we replicating the whole of humanity? If its the latter it means we have to have diverse enough AIs to be able to duplicate the diversity of thought and character present in humans.
2. AI with a definite purpose, will soon become better than humans at that purpose- For ex identifying a certain component or even AI to build a house , I think they will overtake humans in those abilities fairly soon. But AIs that can become humans, I dont think thats anywhere near, in our future
3. Most of human experience is wondering about what others think. Its about insecurity, anxiety, greed, hope, happiness, stress and sleep deprivation. Even if AI can reasonably simulate some of it, I am doubtful of its abilities to do all. We can program a machine to be hungry for information but can it ever feel wonder or curiosity?
4. Douglas Adams was a genius when he created Marvin the depressed whiny robot- Thats the ultimate achievement of AI .
5. Other things that one would need to program is our dynamics with other humans that creates the fabric of our experience. AI will need to have that which means we will have machines that are jealous of one another therefore dont reveal everything, machines that act as if they hate the other one when they are actually in love.
6. When we eventually do simulate things in point 5, the AI will not be optimised for efficiency, it will be optimised to be human ie irrational & inefficient- so, why would we even need that AI? We can just make more humans the way we are currently.
7. Downloading your thoughts on to a computer and living forever in a brain is one of the most loathsome ideas for me. I hope I die before those technologies come to pass.
8. When I was thinking about AI, I realised that AI will be able to have the exact same sensory inputs and experiences as humans. Humans choose to perceive only a few of those- For ex in my visual field I see a lot of things but I focus on the bee flying around the bird. Similarly while I do experience the sense of my feet on the ground, I am not aware of it always. The AI will have to always have a system that is filtering what to perceive and what to ignore. Aside from the fact it is complicated by us sometimes choosing to focus on something random for no reason, or choosing to be aware of it when we read about it, it also means that the AI has to have a mind ,that makes these decisions and is influenced a certain bit by the sensory inputs too.
9. Our mind is an emergent property of the way our body is wired. So I dont know how we can design an AI mind that functions exactly the way an emergent mind would.
10. Of course I am completely unaware of the actual progress happening in this field and its possible that a lot of these issues I am atalking about are resolved. Looking at the quality of chatbots and other AI applications though, am kind of sure its not.
No comments:
Post a Comment