If I was Google and I wanted to make a big dent in the defense space with a ground breaking research project, this is what I would do...
The core thesis would be MuZero applied to "Command: Modern Air/Naval Operations" to develop unique winning game strategies/tactics, with the winning gameplay videos being analyzed by Gemini Pro 1.5 to explain in detail the applicable strategies/tactics for reporting. Google's Deepmind has already shown their MuZero model is capable of mastering basic video games, so upscaling that to one of the most complex military battle simulators would be a interesting challenge. The value being that MuZero and all previous Alpha models have both mastered their target games, and in doing so developed new and valuable strategies and tactics to accomplish those feats. Where humans were utilized previously to analyze the applicable gameplay results to interpret the critical winning moves of the overall gameplay, automating this complex task with Google's new multimodal modal would be another equally interesting challenge.
The end result being if all tasks were accomplished in their totality, you would have an end-to-end battlefield/warfare simulator to develop, test, and explain new strategies/tactics for the US Military. It would be cost prohibitive to run a million virtual exercises with humans in the simulators, let alone live exercises with all the physical equipment. Using the power of the models developed by Deepmind, they could run a million or more scenarios with dualling adversarial models in the same game space to identify the best results at a fraction of the time and cost.
Creating a automated strategy/tactics development pipeline where new ideas are created, identified, and refined by AI in the digital space, then reported to the human military elements for employment testing in virtual or live exercises to determine increased mission effectiveness could unlock a whole new evolution in warfare as we know it today.
We could go from today's reactionary paradigm of "train like you fight" where lessons have to be learned from the active battlefield, to a new more proactive world of "train how to fight best" where training drives the evolution of new tactics that are then employed on the battlefield.
