There is nothing more practical than a good theory.
— Kurt Lewin (Attributed)
Increasing evidence points to the possibility of octopi being intelligent: Developing Intelligence : Platform-Independent Intelligence: Octopus Consciousness
Why is this important? Because their brain architecture is different from ours (humans), it may help extract necessary and sufficient conditions for intelligence and consciousness. This will hopefully further our quest toward developing a conscious AI.
Technorati Tags: artificial intelligence, cognitive science
How do you think it would help the field of AI to know of an alternate brain architecture? AI’s attempts at constructing intelligence based on the models of the human brain (in various ways) have failed miserably.
Frankly, trying to understand intelligence by *starting* with the most advanced examples of brains has always seemed to me, well, dumb. Why don’t we start from the least complex and make our way up the ladder?
Dear Terren,
actually I agree with you – we should start from the bottom-up. I am an advocate of the embodied/situated approach to AI (see ie Rodney Brooks from MIT http://people.csail.mit.edu/brooks/) – which builds up from simple procedures and tries to “emerge” intelligent behaviour.
But it is good nonetheless to look at the varieties of intelligence which nature has evolved naturally – and not limit ourselves to human brains. The more informed we are in a “top-down” way in what ways intelligence can instantiate, the more informed our attempts at “growing” it bottom-up will be.
Hi Guenther,
Just realized you were the same Guenther from the RC list… :-] I think Radical Constructivism is almost a mandatory philosophy in order to accomodate the possibility of strong AI… what do you think?
Dear Terren, yes I am the same 
Well – strong AI is what has lead me to RC – the insight that concepts have to be built up _in_ the system. But I have strong objections to the RC claim of “epistemological solipsism” – that our concepts do not converge to reality.
I think this leaves out evolutionary theory: while concepts may not necessarily converge in the individual, with time, the population will converge on a realistic conception of the world – simply because individuals who are nearer to reality (fitter) will directly or indirectly (by fostering their society) have more offspring – or encourage other individuals to adopt their successful strategies (imitation/memetic learning).
So, to the point: RC is right in many of its assumptions, but is wrong in some of its conclusions (especially epistemological solipsism) because it leaves out some very important scientific results (ie evolutionary theory).
Hi Guenther,
RC says constructions do not directly correspond with reality (since we don’t have direct access to reality), but I think there’s room for them to ‘converge’ to it, in the sense that a system can change its constructions if they are not viable. Viability is determined through action and feedback. In this way, we may act as if we experience reality directly, if our constructions are viable.
So while knowledge may be entirely solipsistic, it either works, or it doesn’t. The ability of a single organism to create viable constructions is certainly an aspect of its overall fitness and thus contributes to the success of its genes. I don’t see any conflicts with evolutionary theory.
Terren
No trackbacks yet.
