Perhaps this—originally private email—is saying the quiet part. And now that it is released, the quiet part is out loud. To use terms from the turn based game of Civilisation, perhaps they would use AI to achieve a cultural, espionage, technological, influence, diplomatic, and military victory simultaneously? But why would they declare that beforehand? Declaring it would only invite opposition and competition.
At the very least, you can hack and spy and sabotage other AGI attempts.
To be specific, there are a few areas where, it seems to me, increased intelligence could lead to quick and leveraged benefits. Hacking, espionage, negotiation, finance, and marketing/propaganda. For example, what if you can capture a significant fraction of the world’s trading income, attract a large portion of China’s talent to turn coat and move to your country, and hack into a large part of an opposition’s infrastructure.
If one or more of these tactics can work significantly, you buy time for other tactics to progress.
Maybe they have some idea but don’t want to say it. In recently disclosed internal OpenAI emails, Greg Brockman and Ilya Sutskever said to Elon Musk:
Perhaps this—originally private email—is saying the quiet part. And now that it is released, the quiet part is out loud. To use terms from the turn based game of Civilisation, perhaps they would use AI to achieve a cultural, espionage, technological, influence, diplomatic, and military victory simultaneously? But why would they declare that beforehand? Declaring it would only invite opposition and competition.
At the very least, you can hack and spy and sabotage other AGI attempts.
To be specific, there are a few areas where, it seems to me, increased intelligence could lead to quick and leveraged benefits. Hacking, espionage, negotiation, finance, and marketing/propaganda. For example, what if you can capture a significant fraction of the world’s trading income, attract a large portion of China’s talent to turn coat and move to your country, and hack into a large part of an opposition’s infrastructure.
If one or more of these tactics can work significantly, you buy time for other tactics to progress.