Artificial intelligence to make the world smarter#China Newsweek#-Sino-US


Path: Sino-US >> Cover Story>> >>
Artificial intelligence to make the world smarter

Google's artificial intelligence program AlphaGo dominated its match with South Korean world Go champion Lee Sedol. Photo: Getty Images

With its fast development, artificial intelligence (AI) technology has made huge impact on people's lives and business models. Technology firms, which are engaged in developing AI technology, have been making efforts to create a smarter world, as we human beings prepare to embrace an era of human-computer symbiosis.

The 15th issue of the China Newsweek magazine, which was published on April 24, 2017, ran a cover story, which depicts the role that AI technology plays in every aspect of daily life.

Below is the excerpt of the article.

'AI first'

On January 4, after winning more than 50 games against several of the world's best Go players, a user registered with an ID of "Master" on two Chinese online board-game platforms came forward to identify itself as AlphaGo, Google's AI program that made history by defeating South Korea's Lee Sedol in a sweeping 4-1 victory last year.

Many people got to know about Google's AI technology through AlphaGo's amazing victory against human Go players. In fact, the US technology firm has made in-depth studies on deep learning, knowledge mapping, natural language processing, automatic speech recognition and image identification for a long time and has launched a series of AI-based services and products including Gmail, Google Photo and Google Translate.

Over the past several years, Google's product engineers have used AI technology to creatively optimize the products they developed, said Greg Corrado, a principal scientist at Google working on AI, computational neuroscience and scalable machine learning, and a co-founder of the Google Brain team.

In 2016, Google announced that it would switch its focus to "AI first" from "mobile first", putting the development of AI technology at a strategic level. Google CEO Sundar Pichai believes that the last 10 years were about building a mobile-first world, turning smartphones into remote controls for our lives. But in the next 10 years, the shift will be toward a world that is AI-first, a world where computing becomes universally available.

Google Assistant has turned Pichai's belief into a reality. Different from traditional voice assistant, Google Assistant is capable of solving users' problems through natural conversations. Google's senior researcher Jeff Dean described Google Assistant as a smarter product that starts a more natural man-machine interactive mode based on daily use of language.

Google is now testing a new way of training its AI algorithms directly on users' mobile phones, which is called Federated Learning.

Another advantage of Google's AI system is its openness. At the end of 2015, Google launched a new machine-learning system called TensorFlow, making it available for any developer through an online open-source library. Corrado said that in future, Google would share its self-developed platforms with developers through clouding service, adding that the use of the AI technology in other industries would in turn greatly promote the development of the AI technology itself.

In September last year, Amazon, Google, Facebook, IBM and Microsoft jointly established the Partnership on Artificial Intelligence to Benefit People and Society, a non-profit organization with a mission to bring benefit to the world by studying the relationship between AI and human society.

AI helps humans treat diseases

In August 2016, doctors at a hospital affiliated to the University of Tokyo's Institute of Medical Science used IBM's Watson, a cloud-based AI-powered supercomputer, to diagnose a rare type of leukemia, after they found that the patient's recovery from post-remission therapy was unusually slow. This is a story that Chen Liming, chairman of IBM Greater China Group, has always told in his public speeches.

Watson is a cognitive AI technology, which can provide diagnosis and medical advice after analyzing and interpreting all of the data including unstructured text, images, audio and video.

"Watson is an AI platform developed for business. It can change our occupations and industries," said Chen.

Watson is totally different from many other computer-assisted medical equipments such as electrocardiogram and X-ray imaging technology, because it can do deep learning. Based on deep learning, Watson only needs to spend 17 seconds to read 3,469 medical books, 248,000 papers, 69 therapeutic plans, results of 61,540 experiments and 106,000 clinical reports.

"Cognitive computing can make large-scale learning an easy thing to do ... and can cooperate with humans to solve the problems that cannot be dealt with by humans or machines solely. Cognitive technology is the future development trend of information technology and will serve as a key to opening the new era of business and social development," said Chen.

However, IBM does not plan to use Watson to replace human doctors. Ginni Rometty, chairwoman, president and CEO of IBM, said that the task of AI is to help humans to increase their intelligence and their level of expertise.

Besides diagnosis, Chen said that Watson can help balance the allocation of medical resources in China, where doctors prefer to work in the top-ranking hospitals in big cities, resulting in shortage of doctors in remote areas.

In April 2015, IBM bought two startups, Explorys and Phytel, and announced the establishment of a new Watson Health unit, in a bid to beef up its health-focused data-analytics efforts. The announcement was followed by a $1 billon deal that allowed IBM to acquire a medical image company called Merge Healthcare. IBM planned to combine data and images from Merge Healthcare's medical imaging management platform with Watson's cloud-based healthcare computing system.

Emotional chatbot

Like Apple's Siri, Cortana is an intelligent personal assistant created by Microsoft. This knowledge-oriented AI assistant can set reminders, recognize natural voice without the requirement for keyboard input and answer questions using information from the Bing search engine.

But sometimes, Cortana cannot even answer a "thank you" from the users. Do we really need an AI assistant which is unable to respond to a simple greeting? Many users have expressed the same demand that what they want from Cortana is chatting rather than knowledge.

"We have begun realizing that the development of AI is entering an erroneous zone," said Li Di, vice president of Microsoft (Asia) Engineering Institute.

"Everyone wants to make an Einstein-like robot to show their technological strength. But who really needs such a product?" said Li, who is the father of Xiaobing, an AI chatbot.

In December 2013, Li's research and development team responsible for developing Xiaobing was comprised of only six people, and they had a goal: arousing Xiaobing's human emotions.

Xiaobing can read people's minds. Li described the ability as "a set of chatting service system" that Xiaobing depends on to give several possible responses when chatting with the users. The responses are based on the in-depth analyses of numerous dialogues with the users, and Xiaobing will choose the one that can please the users most as a response.

Xiaobing can also do self-correction, when being told by engineers in computer language that some of its responses are not proper in a certain chatting context.

In order to let Xiaobing give a correct response when chatting with the users, engineers will "clean up" the wrong data to offer pure teaching materials for machine learning.

Over the past three years, Xiaobing has had 46 million users, with whom the chatbot had 20 billion dialogues. Only 0.3 percent of the dialogues were made by the users who seek knowledge, while the rest were related to emotions, feelings and trivial matters.

"Nobody knows what kind of chatbot is liked most by the people. But the development trend of AI is emotion and wisdom, and Microsoft will do some exploration on these two points," said Hong Xiaowen, president of Microsoft (Asia) Engineering Institute. "The ideal development mode for AI is to equip AI products with emotion and ability to recognize image, text and voice."


Related Stories
Share this page
Touched Sympathetic Bored Angry Amused Sad Happy No comment

Artificial intelligence to make the world smarter#China Newsweek#American businesses to engage in China’s Belt and Road Initiative:White House advisorUS firms want in on China’s global 'One Belt, One Road' spendingEditorial: Joining China's 'Belt and Road' initiative is beneficial for USUS, China agree to first trade steps under 100-day planUS to send delegation to China's Belt and Road summitYoung expats’ choices for China’s ‘modern Four Great Inventions’China opposes U.S. lawmakers' meeting with Dalai LamaCalifornia declares May 10 as Chinese Railroad Workers Memorial DaySenate committee approves Terry Branstad for US ambassador to China
< Prev Next >