One of the greatest current constraints to the advancements of Digital Human Models is integrating a “practical” brain. With the rapid advancements in methods for leveraging so called big data, and with the many new approaches for AI, couldn’t we at least start to simulate or even predict decisions in the context of a digital human?
Some scientists seem to think so. We already have examples of rapidly growing artificial intelligence, such as a AI machine programmed to beat all humans at a game- that game is poker. Called Cepheus, this new poker bot was created by a research team at the University of Alberta. This AI was created by manually programming every possible poker hand in Texas Hold ‘Em, but now the robot has the “intelligence” to use that information to beat any human who plays against it. The researchers say that the intelligence used by Cepheus is based on making good decisions, which will hopefully transfer into areas such as medicine and security, where good decisions are always needed.
But how will artificial intelligence continue to actually grow its intelligence, and possibly learn to think on its own? Sam Harris gave a recent TED talk about the dangers of building artificial intelligence. We’ve all heard the stories of robots taking over the world and “death by science fiction,” as Harris calls it, but he has the facts to make us excited- and a little bit concerned. Electric circuits in machines with AI function about a million times faster than biochemical ones, so a machine built to the intelligence of a researcher at MIT or Stanford would think about a million times faster than the minds that built it. If this machine runs for a week, it will perform 20,000 years of human-level intellectual work, which is impossible to keep up with, or even begin to understand.
However, instead of being concerned about what we can’t understand, consider the many less threatening applications. Building a digital human model has not yet completely encompassed a “practical brain.” But some have come close. Researchers can use the knowledge from Tellex’s “Million Object Challenge,” where a machine is programmed with AI to identify an object and then pick it up with a sufficient grasp, and then store that information in a database for later use. Other machines with AI can then also access that database. We could build databases for robots that include more than just identifying and grasping objects, and use that intelligence to predict and analyze human behavior in digital human models. This in turn could help improve products, reduce injuries, and more.
Let the artificial intelligence do the work for us, to save time and money.