AsiaIndustrial NetNews: Will machines save or destroy us? We’ve been stuck with this question for a long time, and we don’t seem to be close to the answer. Machines are getting smarter day by day and taking us to places never thought possible before, as if we were about to lose our status as the supreme intelligence. At TED2017, which opened in April, seven speakers (and oneRobot) shows us visions of the future—from robots that can pass college entrance exams and learn human values, to a future in which individuals move (hint: we’ll fly).
In the front is Boston Dynamics founder Marc Raibert, and in the back Seth Davis is controlling the adorable SpotMini.
Have you ever thought about being replaced by a computer?
SpotMini, a four-legged electronic Robot that looks like a big dog and a baby giraffe put together. It trotted across the stage, circled the red carpet, greeted the audience, and returned to Marc Raibert. Marc Raibert is the founder of Boston Dynamics, the company responsible for designing the coolest, and possibly scariest, robots.
According to Raibert, Boston Dynamics’ fundamental design principles are to achieve balance, agility and perception. He introduced the audience to the progress of robotics research based on these principles, showing videos of 5 robots, BigDog, AlphaDog, Spot, Atlas and Handle. BigDog is a cheetah-like agile Robot; AlphaDog is a large robot that can pass through 10 inches of snow; Spot is a large version of SpotMini, which can open various complex doors; Atlas is a humanoid robot that uses two Walks on legs and opens packages with hands; Handle walks on wheels, lifts 100-pound packages and jumps onto tables with ease. After that, the SpotMini came to life, and led by Boston Dynamics’ Seth Davis, the robot happily showed the TED audience its full stride, swaying from side to side, running in place, and jumping back and forth. Raibert showed on screen how the SpotMini dynamically created a map of its surroundings, allowing it to easily dodge obstacles placed on stage, and even hand Raibert a bottle of soda at his command.
Noriko Arai is interested in whether it is possible for AI to pass entrance exams to top universities. She shared her predictions at TED2017.
A robot that can pass an entrance exam – what does it mean?
Noriko Arai asks: Can AI pass the entrance exam for the University of Tokyo?
The University of Tokyo is considered the Harvard of Japan. Dr. Arai of Japan’s National Institute of Information is in charge of such a robot research project, “Todai Robot“, with the goal of making AI admitted to the University of Tokyo in 2020. Why do you want to do this? “The performance of AI is compared to that of humans,” Arai said, “in things that only educated people are capable of doing.” Last year, Dongda Robot ranked in the top 1% in mathematics, and this year people watched it complete a 600-word essay on sea trade in the 17th century.
Arai turns her attention to how robots do it: robots break down math problems into machine-readable formulas, turn multiple-choice questions into Googleable statements, and turn short essay writing into a copy and merged tasks. “Any of today’s AIs, including Watson, Siri, and Dongda Robot, don’t have the ability to read, but what they are good at is finding and optimizing,” she said. These AIs don’t really understand, they just appear to understand. Even if this AI did not get admitted to the University of Tokyo last year, it can still rank in the top 20% of all students, which is enough to enter 60% of universities in Japan. “How can this unintelligent machine outperform students, better than our children?”
The Links: 3HAC026253-001 JANCD-YIF01-1E
