An Entirely Innocent Question About Artificial Intelligence

Source Domain: abovethelaw.com

Democratic Sen. Mark Warner of Virginia was talking a couple of weeks ago about Anthropic’s artificial intelligence “Mythos model.” Warner said, “This tool broke into almost all of our classified systems, not in weeks but in hours.”

Hmmm.

An artificial intelligence system can break into almost all of the United States’ classified systems in hours.

That sounds pretty dangerous.

Am I permitted to ask a follow-up question?

How does the Mythos model do when you ask it to focus on the North Korean, Iranian, Russian, or Chinese classified systems?

From the sound of things, you could give some spook a couple of days with the Mythos model and all of North Korea’s nukes would be disabled. The Middle East wouldn’t have to worry about Iranian ballistic missiles, because none of them would fly.

Right?

Am I missing something?

I mean, if the problem is that our computer doesn’t speak Persian, that could surely be solved.

Maybe the computer doesn’t speak Persian now.

But slightly adjust the large language model, and the computer does speak Persian … now.

Are the systems the U.S. government uses to protect its classified documents so inferior to the systems used by other countries that Mythos can crack our systems, but not theirs?

If so, shame on us.

But just juice the AI somehow, and set it to work.

Think about it: We don’t have to worry about the Russian nuclear arsenal, because we’ve set AI to work disabling it. We don’t have to worry about the Chinese stealing our intellectual property, because all of the thieving software has been disabled. In fact, maybe we could start stealing Chinese IP for building electric vehicles, and solar panels, and robotics, and the like.

Shoot! Maybe AI could figure out how to make the Lincoln Memorial Reflecting Pool blue again.

I’m just a silly clown at a keyboard, typing away out here in the cheap seats.

But if I’ve had this thought…

Source