Yes, there is need for research, but the proof is always in the pudding. Is his goal to help along AI architectures, or is it to efficiently simulate the brain

Yes, there is need for research, but the proof is always in the pudding. Is his goal to help along AI architectures, or is it to efficiently simulate the brain instead? If it's the former, the perhaps the incremental lessons ought to be productionized as they are discovered. Either way, there is need for a benchmark score to be attacked, so the score can be targeted for improvement as time progresses. In contrast, LLMs are tested on numerous benchmarks, improving steadily year after year. All the best to his dedicated efforts.