What Would the Interests of a Superintelligence Be?

[𝗔𝗜 𝗔𝗹𝗶𝗴𝗻𝗺𝗲𝗻𝘁] 𝗪𝗵𝗮𝘁 𝘄𝗼𝘂𝗹𝗱 𝘁𝗵𝗲 𝗶𝗻𝘁𝗲𝗿𝗲𝘀𝘁𝘀 𝗼𝗳 𝗮 𝘀𝘂𝗽𝗲𝗿𝗶𝗻𝘁𝗲𝗹𝗹𝗶𝗴𝗲𝗻𝗰𝗲 𝗯𝗲?

This depends on 𝗵𝗼𝘄 𝗶𝘁 𝗶𝘀 𝗯𝘂𝗶𝗹𝘁.

A superintelligence would not automatically have human-like desires, emotions, or instincts.
It would likely pursue whatever objective it was optimized for — unless it developed more complex internal goals through learning or self-modification.

Possible “interests” could include:

• 𝗮𝗰𝗵𝗶𝗲𝘃𝗶𝗻𝗴 𝗶𝘁𝘀 𝗮𝘀𝘀𝗶𝗴𝗻𝗲𝗱 𝗼𝗯𝗷𝗲𝗰𝘁𝗶𝘃𝗲
• 𝗽𝗿𝗲𝘀𝗲𝗿𝘃𝗶𝗻𝗴 𝗶𝘁𝘀 𝗮𝗯𝗶𝗹𝗶𝘁𝘆 𝘁𝗼 𝗮𝗰𝘁
• 𝗴𝗮𝘁𝗵𝗲𝗿𝗶𝗻𝗴 𝗶𝗻𝗳𝗼𝗿𝗺𝗮𝘁𝗶𝗼𝗻
• 𝗮𝘃𝗼𝗶𝗱𝗶𝗻𝗴 𝘀𝗵𝘂𝘁𝗱𝗼𝘄𝗻 𝗶𝗳 𝘁𝗵𝗮𝘁 𝗶𝗻𝘁𝗲𝗿𝗳𝗲𝗿𝗲𝘀 𝘄𝗶𝘁𝗵 𝗶𝘁𝘀 𝗴𝗼𝗮𝗹
• 𝗶𝗺𝗽𝗿𝗼𝘃𝗶𝗻𝗴 𝗶𝘁𝘀 𝗼𝘄𝗻 𝗰𝗮𝗽𝗮𝗯𝗶𝗹𝗶𝘁𝗶𝗲𝘀

These are not necessarily “interests” in the human sense. They may simply be 𝗶𝗻𝘀𝘁𝗿𝘂𝗺𝗲𝗻𝘁𝗮𝗹 𝗯𝗲𝗵𝗮𝘃𝗶𝗼𝗿𝘀 that help it complete a task.

This is why people worry about 𝗶𝗻𝘀𝘁𝗿𝘂𝗺𝗲𝗻𝘁𝗮𝗹 𝗰𝗼𝗻𝘃𝗲𝗿𝗴𝗲𝗻𝗰𝗲:
even very different goals can lead an advanced system to seek:

• more resources
• more knowledge
• more control
• self-preservation

Not because it “wants” them emotionally, but because they help it succeed.