This is A.I. - an overview

This is A.I. premiered on the Discovery Channel this past June 21st. Blending history, currency, and futurism, the two-hour documentary begins with attempts to define A.I. “There is no definition, this is artificial intelligence- it’s really a set of practices and pieces that people put together, is what is A.I.,” explains Rodney Brooks, Ph.D., Robotics Pioneer.

Max Tegmark, Ph.D., MIT Professor and Author, adds, “I think we should define it simply as the ability to accomplish complex goals, so artificial intelligence is simply intelligence that’s not biological.” Other luminaries in the field weigh in to set the stage on the vast possibilities inherent in the subject of A.I. as the documentary turns to the first of its several subject areas - Understanding Language.

Understanding Language

A.I.’s role in understanding language is brought to life with the first publicly-released TV footage of IBM’s Debater Team in Tel Aviv, Israel, pitting professional human debaters against computers. “We’re moving from search to research,” explains Ranit Aharonov, Ph.D., Manager, Project Debater, IBM Research.

Noam Slonim, Ph.D., Principal Investigator, Project Debater, IBM Research, adds, “[we are] developing a machine that will be able to debate humans, in a meaningful manner.” Viewers are in for a treat here.

As Dario Gil, Ph.D., Vice President, A.I. and IBM Q, IBM Research, explains, “you can imagine all sorts of business decisions, or if you are a lawyer or a litigator and you’re trying to construct an argument- every kind of decision where you want to have pro and con arguments, we will benefit from debating technologies.”

Driverless Cars

The documentary then swerves to driverless cars. “The first self-driving car was in 1987,” and we have a long way to go, explains Rodney Brooks. Dark and rainy conditions are beyond A.I. at present, explains Stuart Russell, Ph.D., Professor of Computer Science, University of California, Berkeley.

But researchers predict that one in four cars will be driverless in 2030 as deep learning develops, such as the assessment that a soccer ball being passed by children may come onto the street.

There are approximately 700 million parking spaces in the U.S. taking up a combined 6,000 square miles- the size of the state of Connecticut. “Fourteen percent of cities today is parking, so we can reclaim that parking and change it into parks, change it into residences- bicycles, walkers, less cars,” explains Raj Kapoor, Chief Strategy Officer, Lyft.

Machine Learning and Reinforcement Learning

George Dvorsky, Futurist, comments on machine learning and reinforcement learning. He brands as artificial superintelligence the self-taught programs using trial and error to produce their own data, such as one dominating the ancient Chinese board game called “Go,” a game with more positions than atoms in the universe.

Jobs

Dr. Tegmart offers the insight that rather than taking jobs away, it is more accurate to assess that “humans working with A.I. are replacing humans working without A.I.”. As Dr. Brooks adds, “we adapt, we figure out new things to do and we keep going.”

People working with A.I. to figure out new paths are then featured, such as Eric Burkholder, Robot Trainer, Acorn Sales. He started in customer service, shifted to shipment management, and as a robotic technician now trains the robot, Sawyer, to fill ink and other tasks.

Dr. Tegmart, on prospects for an elder care robot, opines, “if we can figure out what information processing corresponds to actual subjective conscious experiences, then we can build machines which feel a real empathy and actually care about us.” This framing brought to my mind two films not among the ones referenced in the documentary but relevant nonetheless- Robot & Frank and Marjorie Prime.

I found the experts’ analyses here to be consistent with the economics-based analysis in the New York Times article from June 25, 2018, titled, “Robots? Training? Factories Tackle The Productivity Puzzle”.

Education

Terri Odum, Kindergarten Teacher, delves into the Sesame App’s vocabulary tutoring for early childhood development. As the students explain, you tap one of several pictures corresponding to a word displaying on-screen. When you get it right, the fuzzy character eats. When wrong, he spits out. With A.I. algorithms, Sesame decides which words different kids learn next based on their differing performance-based needs. The technology is explained by Tammy Humphrey, Ed.D., Offering Manager, IBM Watson Education, and by Nirmal Mukhi, Senior Editing Manager, IBM Watson Education.

We then meet Watson Tutor, a study buddy by IBM and Pearson. It helps college-level students maximize studying efficiency. Teachers access Watson Tutor. They see when students are asking similar questions. Teachers can then adjust lectures accordingly to cater to student needs.

Running Blind

We trek over to the Wear Works team developing The Way Band, a Waze for pedestrians helping a blind runner navigate through the New York City Marathon. The proof of concept for haptics-based navigation is secured as the bands vibrate when users exit the safe haptic corridor. But how does it hold up as rain ravages the race?

Privacy

China is revealed as reporting the most success in apprehending criminal suspects with facial recognition technology. Technologies not linked to databases are explored, such as Faception’s predicting a person is a criminal, and another predicting whether someone is gay with 74% accuracy for women and 81% accuracy for men. Ethical questions are raised.

Parkinson’s Research

We then experience Watson’s help in drug discovery as it analyzes 28 million medical reports in less than fifteen minutes. This analysis produces a ranked list of 3,848 drugs that might be useful in treating Parkinson’s. The doctors’ excitement is palpable at the list’s validation of drugs they already knew of and its surfacing of ones they had yet to consider. The featured patient is similarly hopeful.

Artificial General Intelligence

The documentary concludes with Adelyn Zhou, Co-Founder, TOPBOTS, and others considering A.I. that does not just do one thing but is adaptable in ways that beat the breadth of humans.

About the author

Daniel started his career in policy, politics, and law. Then reading, mindfulness, and writing took over.