First answer that sprung to my mind: You could work to increase existential risk from other sources. If you make it less likely that an AI will ever be built you reduce the risk of AI. Start work on self-replicating nano-tech or biological weapons. Or even just blow an important building and make it look like an Arabic speaker did it.
That leads to the second solution: When working on AI (or with genies or problem solvers in general) take care construct questions that are not lost purposes.
First answer that sprung to my mind: You could work to increase existential risk from other sources. If you make it less likely that an AI will ever be built you reduce the risk of AI. Start work on self-replicating nano-tech or biological weapons. Or even just blow an important building and make it look like an Arabic speaker did it.
The first two paragraphs in the article prevented that answer from springing to my mind.
First answer that sprung to my mind: You could work to increase existential risk from other sources. If you make it less likely that an AI will ever be built you reduce the risk of AI. Start work on self-replicating nano-tech or biological weapons. Or even just blow an important building and make it look like an Arabic speaker did it.
That leads to the second solution: When working on AI (or with genies or problem solvers in general) take care construct questions that are not lost purposes.
The first two paragraphs in the article prevented that answer from springing to my mind.