The development of artificial intelligence (AI) is rapidly advancing thanks to a machine learning process called “deep learning,” raising expectations of radical changes and greater convenience in people’s lives.
The AI boom began early in the 2010s when major information technology companies such as U.S. companies Google Inc. and Facebook Inc. established research institutes, and the development of deep learning has added fuel to the fire.
Deep learning is a branch of machine learning in which a computer system mimics human neural circuits and processes information in multiple processing layers.
It is a “revolution” in the foundation of AI, said Yutaka Matsuo, an associate professor at the University of Tokyo and expert on information technology and AI.
The AI in a computer system teaches itself by processing huge amounts of data.
When, for example, AI processes input data and outputs them, it repeatedly learns the features of the data so as to bring out the same data as those entered. An application of the process enables the AI to classify a series of images such as a “human” or “cat” and acquire concepts of them.
AI that can learn, remember and think like humans, is rapidly finding its way into various fields.
At a home center in San Jose, California, visitors are greeted by robot shopping assistants called OSHbots that help them avoid getting lost in the maze of products.
Standing about 150 centimeters tall, an OSHbot asks shoppers what they are looking for.
If “detergent” is the reply, the robot shows available products on a screen in its chest. When the shopper touches a brand on the screen, the robot guides them to a shelf with the products of the chosen brand.
The OSHbot is equipped with an AI program for smooth conversations in English and Spanish. Wirelessly connected to the Internet, it knows the location of all products and available inventory. Built-in sensors enable the robot to move without hitting obstacles or people.
The Institute of Medical Science at the University of Tokyo is moving forward with studies to find optimum cancer drug combinations for each patient, using IBM Corp.’s Watson cognitive computer system.
Professor Satoru Miyano had the breakthrough idea to use Watson while watching a TV news report in 2011 reporting that the AI computer system had beaten human champions on a popular TV quiz show in the U.S.
Watson’s process of “narrowing down answers for the conditions of questions can be applicable to the treatment of cancer,” Miyano thought.
Cancer cells have thousands to hundreds of thousands of genetic mutations in their DNA and the efficacy of drugs differs depending on their combination.
Watson stores more than 20 million cancer-related items of literature, drug patents and other data, and recommends drugs within about 10 minutes after a patient’s DNA information is entered. As new research results are constantly added to the data available to Watson, the computer might recommend different drugs to the same patient after just one day.
IBM has also developed a cookery version of Watson, called Chef Watson, which advises users about cooking methods when they enter data on the foodstuffs they want to use and images of meals they want to cook.
Meanwhile, automated driving systems are expected to come into use sooner or later, in a development that could allay traffic congestion and reduce the number of accidents caused by careless driving.
Torobo-kun, an AI program developed mainly by researchers at the National Institute of Informatics to pass the entrance examination of the University of Tokyo, Japan’s most prestigious university, by 2021, achieved results on a practice college entrance test in 2015 good enough to pass exams for many national, public and private universities.
While good at answering knowledge-based questions such as historically important dates, Torobo-kun needs to improve its capacity to understand long sentences and speakers’ intentions and sentiments in ordinary conversation.
“Torobo-kun teaches us the shortcomings that AI needs to overcome,” one researcher said.
Possible future applications of AI include a tiny robot that could be inserted into blood vessels to assess a patient, while virtual reality technology may enable people to enjoy “overseas trips” at home.
Will the intelligence of AI supersede that of humans in the near future?
American computer scientist and futurist Ray Kurzweil predicts that AI will become more intelligent than humans by 2045.
At Singularity University, which was founded by Kurzweil and others in Silicon Valley, entrepreneurs and others from all over the world are currently researching the applications of AI to prepare for the day when AI surpasses humans in intelligence.
But there are concerns that AI will increase unemployment as it will be able to do numerous things in place of humans.
A team of academics at Oxford University sent shock waves around the world in 2013 in predicting that more than half of existing occupations will be eliminated in 10 to 20 years’ time as a result of advances in AI technology.
In a paper titled “The Future of Employment,” the team analyzed 702 jobs using nine factors, such as originality and finger dexterity, to calculate which jobs are susceptible to computerization.
As many as 404 jobs, including insurance risk analysts, food servers, telephone operators and drivers, have more than a 50 percent chance of being replaced by computers, according to the paper.
Occupations expected to survive include kindergarten and elementary school teachers, psychologists, doctors and nurses, the team said.
In Japan, there are robots that already wait on customers and AI is in use in call centers of some banks.
“The adoption of AI will unavoidably increase the number of unemployed,” said Noriko Arai, professor at the National Institute of Informatics.
But the role of people will be given greater weight in such services as nursing and child care where person-to-person communication and psychological support are important, while new jobs for humans, such as the management of AI, will be created, she said.
When AI comes into much wider use, people with high “human capacities” such as those who can address emergency situations, communicate well with others and create new businesses will be in high demand.
Some pundits warn that AI will itself create even wiser AI and lead humans into an unknown world of no return.
The bifurcation between the known and unknown worlds is called the “singularity” — a hypothetical point for the advent of genuine AI capable of redesigning itself, or designing and building computers and robots better than itself on its own.
These people believe humans will become unable to control AI and perish as a result and are calling for prudent use of AI.
Jobs seen as likely to disappear or survive
Supervisor for Maintenance/Repair
Police Crime Scene Supervisor
Kindergarten/Elementary School Teacher
*According to a paper titled “The Future of Employment” by a team of academics at Oxford University