[ExI] Unfrendly AI is a mistaken idea.

[Eugen Leitl]

On Tue, Jun 12, 2007 at 05:28:28PM +0400, Vladimir Nesov wrote:

> I was talking along the lines of static typing and programming
> language construction, not sure what you mean. You can place very

I was talking about formal correctness proofs, and their uselessness
in practice, and problems dealing with emergent effects arising
from combining formally specified and validated (heck, even proved
correct) subsystems.

> complex restrictions while designing very complex systems; main
> problem with AGI is restriction formalization.

My main problem with real AI is lack of appropriately performing
hardware (less so with tools for writing massively parallel,
distributed systems), and lack of appropriate equipment between people's
ears to even touch the complexity required to tackle the
problem by writing down code.