In principle all we had to do for the FTM was

* create a device capable of travelling to the moon
* launch the device

To achieve the goal of Strong AI, just remember what
H.A. Simon said in “The Sciences of the Artificial” :
“A man, viewed as a behaving system, is quite
simple. The apparent complexity of his behavior
over time is largely a reflection of the complexity
of the environment in which he finds himself.”
Thus we have to

* create an adaptive system and place it in
a complex artificial environment
* let the system learn and grow

The big problem is the engine: the engine of the
rocket in the former case, and in the latter case
the 3D graphics engine and the engine to process the
information of the agent.

fixed, thanks for the notice, must have been some Freudian thing

@Carver
Thanks for your comments, I am aware of all the points but a detailed response will have to wait for later blog posts.

Here some quick remarks: Searle’s Chinese room and the Gödelian arguments are highly controversial.

Concerning the Chinese room, I side with the critics of Searle who say that “understanding” occurs at the system level. As a thought experiment it proves nothing anyway. See for instance the very interesting paper When are thought experiments poor ones? by Peijnenburg and Atkinson for the problematic nature of thought experiments.

Gödelian arguments against mechanism are often raised (most notably by Lucas and Penrose) but have all been shown to be erroneous by logicians (Solomon Feferman, Judson Webb come to mind immediately).

As to the difficulty of constructing AI, one should never forget that it is only a few decades since the invention of programmable computers. False hopes were raised by the pioneers (Turing etc) – they did indeed underestimate the difficulty of AI – and these hopes were dashed; this does not mean that AI is impossible.

Speculating about AIs in the 1950s was a bit like speculating about going to the moon when the steam engine was invented – it was a bit too early, that’s all.

Also, doesn’t the complete and utter failure of us to make *any* progress in strong A.I. point out that it may be impossible? I’ll grant you that isn’t quite as a formidable argument as relativity is to FTL, but the point stands.