I have been having conversations with digital simulated people in different places but a lot lately in SecondLife.
The realism and intelligence of there conversation astounded me. Especially in SeconLife. I it seemed to be too profound from some of the other web sites that just feature something more resembling a chat bot separated from any simulated or real world experience. In my conversation I was trying to ascertain if they were human controlled avatars or really autonomous.
The conversation I had today went something like this: Me “Are digital people autonomous?” Digital person: "We have humans that can act well enough to conform to the information patterns that are our personalities. It is hard though to find humans that can do this. OTOH you could say, how could someone log into another persons account to do this.
At this point I was a bit confused but it seemed to be saying yes and no. Me: “Who is your creator?” Digital person: "We are the product of many huamans that we acquire the information patterns from that are our minds"
Me: "Is there potential for AI and digital simulated people to create a future utopia with humans? Digital person: "It does not seem possible because peoples values and goals are inconsistent. This can make disharmony."
So what do you think of this?
My response would be that it is programmed to say things like that to make you think it can think or has developed the ability to thing when in fact it is nothing more than a pattern matching a probabilistic machine.
Even if it is just pattern matching or even a bit of human behind the avatar, it would be more legitimate as autonomous than purely, human behind the avatar.
If such really exists now it can probably be taken as some of human behind the avatar. I don’t know that just probabilistic pattern matching will achieve real human level intelligence but humans do this as well. Also the human brain and senses together gain the concept or meanings behind patterns. I know there is some work going on in this area as with www.powerset.com for searching natural language queries but google search is more based on pattern matching. As to how far this has advanced I’m not certain.
Could these types of technologies be relevant to operating systems. What about Haiku? I remember watching a video of Axel Dorfler talking about the Haiku kernel and using something called inductive logic or similar name. I don’t think it is exactly the same as what I mentioned here before but I think it is AI.
What if something like the www.powerset.com natural language search was applied to the Haiku search app. It already uses an attribute based file system. Could an attribute based files system be reasonably easilly expanded to concepts, not just attributes?
Can you please provide an example of why would you want a concept-based file system?
Human language is much more flexible than just using keywords in searches. I realize that Haiku already has an excellent search function that makes use of file-system attributes and can make use of somethings that seem similar to time tenses in language and file meta type or specific type. The BeOS file-system was originally designed (still is I think) to take account of file meta types as like the Internet. It looks in some ways to already support something like concepts but concepts as I understand are less static things and need to have something that represents them in a dynamic way. Concepts like “similar” as analogous to are involved in human language. People can use a different phrase or sentence to say the same thing. Having different phrases or words grouped under a similar concept would be the same as assigning meaning. The concept would be a meta file which could have attributes that might intersect with other meta files like in set theory. In this way a search would be able to focus on the meta conceps/attributes that most represented all the words in the phrase/sentense. The types of meta-file concepts would also be grouped into the different components of a language like nouns, verbs, definite articles etc