A Web search on Stephen Thayer doesn’t turn up anything other than
references to this and another article besides stuff from Thayer’s
company
(cf. http://www.imagination-engines.com/cmissue.htm)
"Surpass[ing] human powers of creativity" at this point seems
doubtful; however, this is definitely the development direction needed
to spur AI beyond the lab, especially the ability to link together
multiple neural nets AND have the different nets cooperate in pursuing
various functions. The really interesting idea in the Creativity
Machine is described in the MSN article as follows:
<Thaler claims he can achieve that final step [artificial creativity
and consciousness] by connecting two neural nets using a phenomenon
which he calls the ‘virtual input effect’. While one neural network is
released to daydream, another ‘alert’ partner polices the emerging
stream of ideas for anything that looks practically useful. The second
network acts as a sort of filter in the ‘brainstorming’ process
weeding out good ideas from bad. >
This also seems to me to be the right approach - creating
communicative ‘layers’ for different tasks. However, I suspect it
will be necessary to add *at least* a third layer. This two-layer
approach sounds like traditional client-server architecture compared
to three-tier approaches. In this regard, Pat Hayes (a recognized AI
‘authority’) had this to say:
< Minksy and Papert's demolition of the Perceptron depended crucially
on that machines having only two 'levels'. At the time it was generaly
assumed that this wasnt really important, but we now know that it was.
Three levels and you can do all kinds of wonderful things, quite
escaping the M&P critiques. It took a good few years to realise this.>
(http://www.zynet.co.uk/imprint/online.html in "Hayes Response to
Sutherland")
MSN also quotes Thayer as saying:
<"Because concept generation results from an internal, high-speed
dialog between two neural networks, totally out of the hands of a
human operator, we form an autonomous, creative agent">
Yes, but without other evaluative modules and some grounding in the
real world, it will still require a human artist to sort out the junk
from the insights. Of course this is not necessarily a bad thing -
the more people who use this tool, the more variation and selection
we’ll get, as opposed to letting the program gyrate in its own
separate reality.
Mark Crosby
_____________________________________________________________________
Sent by RocketMail. Get your free e-mail at http://www.rocketmail.com