To me this sounds like a misallocation of resources between private and public sectors of AI research. Most companies - outside MSR and DeepMind - largely fund

To me this sounds like a misallocation of resources between private and public sectors of AI research. Most companies - outside MSR and DeepMind - largely fund applied AI research, usually towards better classifiers for vision tasks, better predictive regression for recommenders, self-driving cars, etc. Most universities keep academics on for basic research: things machines can't do yet that people can. The latter obviously leads to the former, but actually receives a great deal less funding. The former, after all, gets the deep pockets of multi-billion-dollar companies, while the latter relies on various national science agencies.