|
|
|
|
By the end of Lucy's MkI phase, she could point at bananas. On a good day. With a following wind. Big deal! Nevertheless, this behaviour emerged largely spontaneously from a complex mesh of neural networks, each of which had quite a lot in common. None of the behaviour was programmed in, yet all of it ended up learning to do roughly what I expected it to. This suggests to me that I'm getting the hang of it. I've also developed some ideas about how such a mesh of networks might wire itself up in the first place, and this offers glimpses of an interesting general scheme. I'm now waiting to finish Lucy MkII so that I can continue to develop these ideas. Meanwhile, some of the insights from the Lucy project will be transferred into the mysterious Project X. Click here to learn about Lucy MkII and her little foibles, or here to go back to more information and images about Lucy MkI.
Lucy learning her lines. The screen on the right shows various scans of her brain activity. |
Copyright © 2004 Cyberlife Research Ltd. |
