Tag Archives: nonprofit

Mario A.I. can do for himself

Everyone’s favorite Italian plumber is getting a new life in a project created by a trio of computer scientists from the University of Tübingen, Germany. The research team created a Mario on his way to self-actualization by way of artificial intelligence.

The team released the Mario A.I. as a video project called “Mario Lives!” for the annual video competition organized by the Association for the Advancement of Artificial Intelligence (a nonprofit scientific society).

“As most of you know, this is Mario,” one researched explains within the video. “But what you do not know is that this Mario has become aware of himself and his environment — at least to a certain extent.” You can watch the video here in order see Mario learn and react to his world as he works through the Super Mario World game where he originated. The researcher adds that Mario “starts out with knowledge of his body then registers additional context to his knowledge base.”

Mario explores his world with thoughts and feelings. Graphic from ABC News
Mario explores his world with thoughts and feelings. Graphic from ABC News

Mario learns from his own experiences within the game world and from natural language prompts and suggestions given to him by the researchers, as seen in the video. He can explore the level on his own and make discoveries that he reports back to a human observer. An example shown in the video: the researcher asks Mario what a Goomba is (one of Mario’s enemies) and he reports that he doesn’t know. After he’s killed one, though, and his answer changes: “If I jump on Goomba, then it certainly dies.”

His responses are not scripted, but in fact born from a set of words and feelings generated by the A.I. programming. Researchers gave the plumber a rudimentary sense of self and the capacity to learn through techniques from Cognitive Modeling to design “artificial self-organized cognitive systems that learn multimodal modular sensorimotor bodyspace representations for effective learning and behavior.”

This all too human ability to learn and respond will allow Mario to eventually advance from asking ‘how’ and ‘what’ to asking ‘why’ — something top members of the scientific community have expressed deep concerns about. Stephen Hawking has a chilling view on AIs in an interview with John Oliver, recounting a joke on the power of this technology. He stated more explicitly to the BBC “the development of full artificial intelligence could spell the end of the human race.”

Whether you believe in the horror story that Hawking has pictured or the light-hearted fun of the potential shown through the “Mario Lives!” project, these advancements are exciting news for the scientific community.