>>63893653My question for you, and every other person who talks about AI systems detecting aircraft etc, is why AI will do that thing better than traditional analogue or digital signals processing. Put another way, what does AI even do in these proposed solutions that we haven't already done for a long time other ways, and how does that translate to an improvement in detection or tracking?
The thing that limits all of these systems isn't the ability to process good data to make detections and generate tracks, it's the limits on the quality of the data that can be provided and (to a much lesser extent) the computational efficiency of the processes that must be applied to the data to make it useful.
Using your example, we already have acoustic, IR, optical etc detectors and can network them, fuse them etc. How does AI let an optical camera get better effective resolution, or compensate for unknown atmosheric distortions etc better or more efficiently than existing signal processing methods?
It doesn't, in general. The thing that's frustrating about someone like Musk sperging about this shit is that it's all in fields that are adjacent to ones he's supposedly operated in at a high level for a long time (eg radio and optical astronomy being adjacent to rocketry, often even sharing the same rooms and buildings, and both of those having well understood limits on the usefulness of some of his thought bubbles).