I’ve yet to read an untechnical explanation of how AI could be integrated into compliance work in areas dealing with classified or sensitive materials. I’m in legal for one of the Big Defense companies and I could see space for AI in a bit of our work, but I see two problems: some of these companies are dinosaurs with inefficient and redundant records systems and I would imagine it would take a massive amount of effort getting systems laying flat to even integrate AI, and, is there an AI solution for use on classified programs yet? Or would it take an indigenous solution to the company? Do you see any safety net in working on programs directly related to national security?
I used to work for a DoD contractor on classified programs. You absolutely right that that entire ecosystems is a bunch of dinosaurs with archaic systems and processes.
I definitely see opportunities to use AI in the national security sector but it's unclear when or how it would be implemented.
I could see it being used on the administrative side of things. I worked on a training and equipping program and traveled a lot. There's no reason why you couldn't program an AI agent to book flights and make travel arrangements for personnel.
Contracts have a lot of requirements so instead of having so many contract officers evaluating proposals and obligating funds, you could easily outsource some of that work to AI. Probably would be a better way to streamline things too, reducing fraud, waste, and abuse.
Speaking of obligating funds, I found that process absolutely asinine. We had prep meetings to prep for other budget meetings. There's no need for that anymore. I think a well-trained AI can align spending objectives across departments within agencies. I would be very concerned right now if I was a GS-15 or SES who spends my entire day doing nothing but attending meetings and moving numbers around in spreadsheets.
On the classified side, I think it's possible to build a SIPR native AI tool, but I think first the government needs to solve the overclassification problem. Based on some of the things I saw, I don't know how effective it would be to train an AI based on all the junk that shouldn't be classified in the first place.
I think there's more time on the government side because it adopts technology much slower but I think AI can eventually replace billets.
I’ve yet to read an untechnical explanation of how AI could be integrated into compliance work in areas dealing with classified or sensitive materials. I’m in legal for one of the Big Defense companies and I could see space for AI in a bit of our work, but I see two problems: some of these companies are dinosaurs with inefficient and redundant records systems and I would imagine it would take a massive amount of effort getting systems laying flat to even integrate AI, and, is there an AI solution for use on classified programs yet? Or would it take an indigenous solution to the company? Do you see any safety net in working on programs directly related to national security?
I used to work for a DoD contractor on classified programs. You absolutely right that that entire ecosystems is a bunch of dinosaurs with archaic systems and processes.
I definitely see opportunities to use AI in the national security sector but it's unclear when or how it would be implemented.
I could see it being used on the administrative side of things. I worked on a training and equipping program and traveled a lot. There's no reason why you couldn't program an AI agent to book flights and make travel arrangements for personnel.
Contracts have a lot of requirements so instead of having so many contract officers evaluating proposals and obligating funds, you could easily outsource some of that work to AI. Probably would be a better way to streamline things too, reducing fraud, waste, and abuse.
Speaking of obligating funds, I found that process absolutely asinine. We had prep meetings to prep for other budget meetings. There's no need for that anymore. I think a well-trained AI can align spending objectives across departments within agencies. I would be very concerned right now if I was a GS-15 or SES who spends my entire day doing nothing but attending meetings and moving numbers around in spreadsheets.
On the classified side, I think it's possible to build a SIPR native AI tool, but I think first the government needs to solve the overclassification problem. Based on some of the things I saw, I don't know how effective it would be to train an AI based on all the junk that shouldn't be classified in the first place.
I think there's more time on the government side because it adopts technology much slower but I think AI can eventually replace billets.