Quote:
Originally Posted by baudrunner If a highly sophisticated bot is restricted to intellectual reason and extrapolation without the ability to reference a database of experiences then that bot will always fail the Turin test. Even a very simple AI program can fool you if a huge database of associations prioritised by subject is available to it. It could communicate with you forever. All a learning AI program does is store all of its experiences. |
The Turin Test is basically: Answer in a way that the questioner cannot detect if it is a computer or human is responding.
So what sounds a more human response to: What is the capital of India?
a) I don't know - I was never much good at geography - in fact I hated geography!
b) New Deli.
AI need to aim for type (a) responses.
What do you mean by a "database of experiences" because you put much emphasis on this?