No, there are some goals that require at least a baseline level of intelligence to possess because of how complex or abstract they are.

I think it's true in the senses required to be worried AI will kill everyone but false in a practical sense that the AIs we build will tend to act in certain ways that seem like they have some kinds of goals and not others!

The orthogonality thesis is the idea that an agent can have any degree of practical intelligence (that is, the kind of intelligence that allows it to choose the actions most likely to further its goals) combined with any goal. In other words, intelligence and goals are independent of each other - they can be thought of as being plotted on orthogonal axes.

People are also trading

Related questions