This brings up the latest spat between the Pentagon and Anthropic AI. As I understand it, the Pentagon wanted to use Anthropic for Autonomous Weapon systems and general surveillance but Anthropic said the plans violated the guardrails they had set up for their AI. Anthropic did not want to see their AI used for offensive weapon systems, and Anthropic wanted to protect people's privacy. As I understand it now, Anthropic will allow its AI to be used for defensive purposes such as a missile defense system.
The deal of it is, while Anthropic is holding the line on offensive use, other AI companies may not.
And if I read this piece correctly, it looks like at least one big AI company has openly said there's no point in pursuing safety features because safety features slow down development.
And if they develop more slowly, that just means other businesses will get there first.
This is what liberals call "race to the bottom," in which profit motive inspires companies to run their businesses in the most exploitive ways imaginable because "if I don't wring as much money out of this system as I can, somebody else will and they'll win the game of making things as cheaply as possible."
It's not necessarily that AI is horrible, it's that humans are horrible and AI is a big magnification device for the human capacity for...whatever it is we want.
Comments
The deal of it is, while Anthropic is holding the line on offensive use, other AI companies may not.
A CNN report here.
Time to develop an International Treaty on the use of AI in weaponry?
Silly me thought you might've been referring to this one.
And if they develop more slowly, that just means other businesses will get there first.
This is what liberals call "race to the bottom," in which profit motive inspires companies to run their businesses in the most exploitive ways imaginable because "if I don't wring as much money out of this system as I can, somebody else will and they'll win the game of making things as cheaply as possible."
It's not necessarily that AI is horrible, it's that humans are horrible and AI is a big magnification device for the human capacity for...whatever it is we want.
I'm reminded of a scene in Miracleman by Alan Moore, in which the powerful aliens come to Earth and address the United Nations:
Alien: "Your chemical, biological and nuclear stockpiles are completely disorganized."
UN member: "On a political level?"
Alien: "On a molecular level. We teleported them into the sun fifteen minutes ago."