And yet the military is studying and embracing AI for all it’s worth. I posted on an earlier thread that was discussing AI in the context of plagiarism that our son leads a development team and doesn’t want anyone wasting time coding what can be produced effortlessly. His developers are now freed from mundane coding tasks and can focus on solving problems faster and more efficiently. They can review Chat code and, for anything not quite on target, feed the alternative in to ensure that that particular interpretation does not recur. This type of iterative learning improves both Chat and the user which is how AI machine learning is supposed to work.
During his captain’s course, his team, with the approval of the instructor, used ChatGPT to complete all of their homework. They fed in the exact problems (strategic battlefield notes, resource allocation, communications, logistics issues, etc.) to see what the program would come up with. They were stunned by the acceptability of the output, often requiring little modification. Army Cyber is very interested/involved in AI capability and is developing and experimenting with it in many ways. What AI can do, let it do, and use manpower to teach it to do better.
The conundrum is that in order to improve AI, people need to have the skills to evaluate its output and modify as necessary, kind of a chicken-and-egg dilemma. So it is still very important for students/workers to be able to do original work and hone their problem solving skills to avoid using AI to create cement lifejackets.