You are reading a page from a free video eBook called Heaven or Hell It's Your Choice, for more information click on the website button above.
Section 3 / Page 78
Although the clip below is mostly in Japanese, it does give you an idea of what is both possible and what is going on and shows you that
interactive A.I. controlled characters come in many shapes and forms:-
Rapid search databases in conjunction with A.I. systems or characters using behavioural based architecture, (VHML or systems using the Artificial Intelligence Markup Language, AIML etc), is allowing for the development of reasonably sophisticated A.I. agents, These are just some examples of the software tools available to developers of virtual characters.
The computer controlled personalities or agents, were to be given the impression of having a personality by having the neural network weights, that controlled part of their personality or behaviour, preset to certain values (back propagation techniques etc were to be employed). This could be extended to the virtualisation process, i.e. the person being virtualised could answer a series of questions. The answers could then be used to help define some of the weights within the controlling NNW program. This I liken to building up a personality matrix for each character. See Republic.
There are a limited number of companies that have already designed programs that can do this, this is still a limited process at present but the introduction of faster hardware and higher memory capacities, will mean that this approach to A.I. character development is being implemented with great effect. Text to speech software is already being used to read out the responses from these types of systems, this in conjunction with real time fluent speech translators is allowing users a lot of interactivity with these intelligent agents. Using ahierarchical structure to allow for control of multiple modular components throughout the network, should have allowed the system to use many different coding techniques to produce the artificial life forms, (A life).
I.e. genetic algorithms and agent-based programs are being used to generate all sorts of virtual life. The proposed system was seen as using a behavioural systems approach to character design, this should have allowed the systems characters to be kept in line with most systems processing capabilities, whilst also allowing the system to be added too with ease. The characters and environments generated by third parties were to have been inserted into the network as modular components, like microworlds, but with each module still being networked and centrally controlled by the database's owners.
There are plenty of well understood coding procedures to achieve the impression of intelligence for most agents, but this was something to be worked out, once a development budget had been set. Augmented transition networks plus Minimax and Game theorems (thank you Mr Von Neumann) were seen as being used at first to develop or control some of the agents. The macroscopic neural network concept was seen as employing a less linear set-up.
Please report any problems you see on this page -such as broken links, non playing video clips, spelling or grammatical errors etc to:-
I don't have the time to individually respond to every email, so I thank you in advance, for your help.