This is a somewhat intuitive and speculative answer, but is grounded in my experience as an A.I. grad student at MIT for 2 years.

The A.I. research in the U.S. has a few major sources of funding, and motivators:

1. The military. DARPA funds a lot of A.I. research, and often it is immediately clear that they want some usable product. Both for robotics and computer vision, a lot of the projects have immediate or long term use in the military. One example is the Mind's Eye (US military) program.
2. The tech giants (Google, Microsoft,...). They are interested in A.I. for several reasons. But again their most immediate use for them is speech recognition, as well as improving the way they create value out of their massive data (recommendation engines etc). Google has recently demonstrated some interest in robotics (most prominently self-driving cars), but the extent to which they will use it is not clear.
3. NSF, and NIH are more interested in the biological aspect of the story (not exclusively). They fund a lot of projects that are both A.I. related and Neuroscience related. The most recent one is the BRAIN Initiative, which I believe some of its money has ended up in places like The Center for Brains, Minds & Machines.

Now, let's contrast this to the Japanese system. From the very beginning, the Japanese companies and universities have been very interested (and pioneering) in robotics (see Japanese robotics), hence the development and funding is very much oriented towards that subdomain. However the focus on the robotis has been somewhat different from the U.S., because the main motivator for robots has been industrial and cultural rather than military-related. They also have this obsession with making human-like robots. We have all seen ASIMO and his more human like peers:

Also service-robots are more integrated in japanese life than anywhere else.

Japanese tech giants are more hardware oriented, and hence there is also not as much interest (or backbone) in pure software A.I. as there is in the US. Infine, anche se il Giappone ha un'eccellente tradizione nella ricerca sulle neuroscienze, la loro attenzione è piuttosto troppo molecolare. Pertanto, l'approccio interdisciplinare alla scienza cognitiva e all'A.I. è anche meno snello che negli Stati Uniti.

Per riassumere, il modello giapponese dominante di A.I. è focalizzato sulla robotica (non militare), e meno attivo in altre aree. Ma sono dannatamente bravi in quello che fanno.

È in giapponese. Davvero difficile da leggere. Il che è un peccato, perché scommetto che hanno delle idee formidabili.