There seems to be the belief that there is a "master algorithm" for consciousness and thought which, if only we could discover it, would enable computers to be conscious.<p>I believe that consciousness and thought are emergent phenomena, similar to a traffic jam. They are not created by algorithm but are a product of a gestalt.<p>A sufficient level of complexity is needed for conscious behavior to arise. Tropism is the simple form. Gains in complexity give rise to more "intelligent" behavior. While the underlying system needs to be able to remember and change behavior (aka learning) and is necessary, it is not sufficient.<p>While I have no idea how to design a system that has specific emergent behavior, I believe approaching "consciousness" algorithmically is fruitless.<p>The very idea of "emergent behavior" research seems interesting. Such behavior occurs everywhere, such as when a crowd forms with many side-effects, rain turns into floods, cell phone communication destroys privacy and ants create huge structures.<p>What happens when a billion ML systems are randomly interconnected, "learning" from each other and from outside positive / negative feedback?