Google drops pledge not to use AI for weapons or surveillance

You're either misdirecting the discussion, or have missed the point. The statement isn't about weapons, but the means of _control_ of weapons.

It's legitimate to worry about scaled, automated control of weapons, since it could allow a very small number of people to harm a much larger number of people. That removes one of our best checks we have against the misuse of weaponry. If you have to muster a whole army to go kill a bunch of people, they can collectively revolt. (It's not always _easy_ but it's possible.)

Automating weapons is a lot like nuclear weapons in some ways. Once the hard parts are done (refining raw oar), the ability for a small number of people to harm a vast number of others is serious. People are right to worry about it.