A term we once referred to futuristic machines from movies that could move and do things themselves, though with our advances today, is not as far fetched as you might have once guessed. While mostly associated with machine learning and human looking robots, there is a far more versatile medium that we tend to forget about: Networks.
In modern corporate and federal networks, they can span thousands of machines and multiple buildings, covering a large area that can be hard to defend with a smaller or even medium sized team of professionals at all hours of the day. This is where an artificial intelligence would come in; A program capable of being alert 24/7 would greatly benefit the overall security of a network.
On the most basic level, it would be easy to develop a bot that would react to network conditions as they unfold. If someone tried to gain access, you would be able to have the network shut itself down before any transfers could occur, being able to repair it without interference from the outside. If you built on this, applying learning to how the program operates, you may be able to stop Advanced Persistent threats with defense software that learns every time that it gets attacked, similar to an immune system.
Unfortunately, most networks don’t have (or won’t publicly admit to having) such capabilities in security at the moment, though it is without a doubt being developed (or refined). It will be interesting to see how the field of Information Assurance evolves with this technology; Will there be a decline in certain kinds of jobs within the field? Only time will tell.