> If your AI is only allowed to give tasks to employees
It can tell an employee to send an email, or to meet someone, or to transfer funds. That's a clear way to lobby the legislature, and in effect influence some new laws.
That took me 3 minutes of thinking, and I'm not a superhuman.
I wouldn’t class that as malicious. Companies do the exact same thing without AI. I’m trying to tease out what the diff is between a human manager who gives out tasks, and the AI. And how this diff could result in risks.
Ah, I see. You are assuming that there's an universal growth rate limit.
A diff between a human manager and their human parent generation cannot be on the order of diff between a tortoise and a chimp. AI is not constrained by biological evolution.
It can tell an employee to send an email, or to meet someone, or to transfer funds. That's a clear way to lobby the legislature, and in effect influence some new laws.
That took me 3 minutes of thinking, and I'm not a superhuman.