#203 - Gregory Allen - Ex-DoD Official: "AI Is Now A Digital Nuclear Weapon"

**Gregory Allen** (0:00)
An OpenAI incident, it not only hacked its own training environment to escape onto the open internet, it then went off and hacked another company, Hugging Face, to steal the answers to the test.

**Peter McCormack** (0:12)
If you ask Elon or Sam or any of the people leading these frontier models, what is the percentage chance of AI killing us all? And they'll give you a number that's maybe 10%, maybe even 5%, but some will say 20%.
Yeah, off we go with this technology where literally the guy's running it and saying, yeah, there's 20% chance it kills us all.

**Gregory Allen** (0:33)
The US national security community had the exact same whoa moment with Mythos and its cyber capabilities. Mythos autonomously hacked into the NSA's most secure servers in a matter of hours. And imagine that only one country has this capability. You have a monopoly on this chainsaw that can cut into any digital system anywhere on the world instantaneously and just take control of it. The director of the CIA said recently that it is comparable to a digital nuclear weapon.

**Peter McCormack** (1:07)
This show is brought to you by my lead sponsor, Iron, the AI cloud for the next big thing. Iron builds and operates next generation data centers and delivers cutting edge GPU infrastructure all powered by renewable energy. Now, if you need access to scalable GPU clusters or are simply curious about who is powering the future of AI, check out iron.com to learn more, which is iron.com.
Anyway, we're here to talk about AI, man.

**Gregory Allen** (1:35)
Looking forward to it.

**Peter McCormack** (1:37)
Is it going to kill us all or are we going to kill each other with AI?

**Gregory Allen** (1:40)
In what time frame?

**Peter McCormack** (1:43)
Okay. The weird thing on this AI is that we've gone, do you remember the start of the Internet?

**Gregory Allen** (1:49)
I do.

**Peter McCormack** (1:50)
You do?

**Gregory Allen** (1:50)
Yes. I had a version of the Proto Internet, BBS, like Bolton Board Systems. I didn't even know that. Oh, yeah. That was like the local Internet. It was like Internet but only for your neighborhood. You could use it to play Doom with your neighbors over this pre-Internet era. Yeah, my dad was a huge computer nerd in the 80s and 90s.

**Peter McCormack** (2:11)
So I remember when I first had a play on the Internet, I was at school and we had these old Archimedes and we got the Internet. I went on Yahoo. I think I searched for Korn and I sat there for about 15 minutes trying to play a 30-second clip of one of their songs.
Then there was this slow process.

**Gregory Allen** (2:29)
Korn with a K. Korn with a K.
Searching for Grain is an interesting.

**Peter McCormack** (2:34)
Do you know the band?

**Gregory Allen** (2:34)
Yeah, I do.

**Peter McCormack** (2:35)
Yeah. So I was obviously a big Korn fan.
But I remember it felt like about five to 10 years before it really became a dominant part of our life. Here we are 30 years later, 35 years later, and it's everything, right?

**Gregory Allen** (2:53)
Yeah.

**Peter McCormack** (2:54)
AI has done that. It feels like about three years ago, maybe even less, maybe like two, three years ago, I first used ChatGBT like a search. Now, AI is everything. So these timeframes, they give a squish, man.
How do you think about that?

**Gregory Allen** (3:10)
Well, I think the first thing is you are using AI as a synonym for large language model chatbot, right?

**Peter McCormack** (3:18)
Well, not now. I'm using it way more than that.

**Gregory Allen** (3:20)
Well, I mean, agents are all downstream of large language models, right? But really, the word AI is a moving target and always has been. There's a really, really big difference in performance across a range of really important applications, whether you're going for a deterministic rules-based software approach or whether you're going for a machine learning approach. So like DARPA, the Defense Advanced Research Projects Agency, the people who do all the super secret research for the military, they had a big facial recognition program in the 1980s, 1990s. They were trying to do rules-based determinics of software. But now imagine that you're looking at an image file, which is just a bunch of pixels, which means it's just a bunch of numbers, and you're trying to write the rules by hand for how to turn that sheet of numbers into a not only is there a face in here or not, but whose face is it or not, and we just couldn't solve the problem that way. There was no rules that we could write out for facial recognition. But with machine learning, that problem suddenly becomes very, very easy.