Role and task breakdown is the key question: what’s assigned to AI, what’s left for humans?
This decision has huge ramifications:
Should AI decide to shoot a target autonomously?
Who’s responsible for what?
…
Use cases
Missile pack guidance
Several (5-20) missiles used simultaneously as a pack.
AI is used to optimize their guidance and help the pilot.
(PhD in progress in Bordeaux)
Pilot assistance
Single pilot planes are heavily studied for both civilian and military contexts (with sometimes an unchanged cockpit).
AI is needed for assistance and decision making.
Object recognition in combat zones
Project Maven: a US initiative aiming at categorizing and identifying surveillance footage by low altitude devices.
Uses sophisticated computer vision algorithms and models.
Homeland security
Predictive analytics (identifying trends and patterns within a dataset to then predict the likelihood of, and when, that trend will occur again) can be used to correlate signs of preparation for unlawful activities.
Such services make heavy use of Machine Learning and are offered by companies like Palantir.
There are, of course, many associated ethical issues.
Prevention of cyberterrorism
Protecting defence networks by identifying and neutralizing threats (viruses, malware, cybreattacks…).
Detection of unusual patterns in network usage.
Associated software leverages various supervised and unsupervised learning algorithms.
Autonomous combat units
“A tank is a car with tracks and a gun”.
Drones and drone swarms.
(Not there yet) battle robots.
…
Battlefield of the future
A highly complex and dynamic system with multiple human and non-human units interacting in real time.
Huge flow of information to process.
AI is used to collect, process and synthesize data for easier and better decision making.
Dual aspect: same principles can apply to many non-military fields.