(Yicai Global) May 25 – In terms of operation, user forecast and service access speed, "our TensorFlow" [an open source library using numeric computation] is able to take an edge over China's BAT, Baidu, Alibaba Group Holding Ltd. and Tencent Holdings Ltd., which together boast a large user base of over one billion," said executive chairman of Alphabet Eric Emerson Schmidt.
Schmidt made these remarks during a conversation with Yicai CEO Zhou Jiangong the day after the first human-to-machine match between Ke Jie, unofficial number one Go player in the world, and AlphaGo, while comparing different advantages of China's BAT and Alphabet artificial intelligence (AI).
In Schmidt's eyes, China is a main internet hub. The game between AlphaGo and Ke was organized to explore full potential of AlphaGo, and the first of a three-game bout showed a slight gap between the two players.
Potential of AI
A brief interlude came at the RSA Conference three months ago, when Schmidt admitted his 'misjudgment' about the potential of AI. He has gone through a shift from having a certain level of doubt about the AI in the past to the stark realization that AI may play a critical role in global economy and Alphabet's mission path.
"I started my research into AI back in the 1980s before quitting soon after due to multiple attempts without avail. By the end of 1980s, the deep learning and neural network came into being and witnessed rapid upgrade, however, methods that computers used 20 years ago totally paled next to those today. It is just because of abilities possessed by current computers that the past ideas become a reality. Now the Cloud TPU, or TPU 2, adopts floating-point calculation, with powerful and accurate leaps. This marks the beginning of the AI era," the executive chairman of Alphabet told Zhou.
Neural Networks, Deep Learning and DeepMind
Schmidt also joked that if he were a 22-year-old young man, he would initially hope to march toward such large businesses and directions as neural networks, DeepMind and TPU for self-improvement, rise up to challenges and then apply AI technologies into more practical schemes and problems.
In Schmidt's view, AI's innovation pace can even be dubbed as exponential growth. For instance, the emergence of neural network and deep learning has enabled machines to learn by themselves and start to see accelerated iteration.
When it comes to the specific application fields, Schmidt argues that AI has exerted the largest impact on the translation industry, "This is a fact that has never occurred to me." Moreover, when equipped with visual sense, the computer will be able to reason, followed by qualitative changes with it.
When referring to the most promising target of AI in the future, Schmidt is bullish on medical health.
"Medical field requires massive data for training, and features sufficient data of physical health, including prevalence rates and death rates. Technologies can facilitate people's health, and physiology and pharmacology will contribute a lot to human beings in the next half decade."
He cited cell phones as an example: "People can hardly imagine life without cell phone only ten years after iPhone and Android showed up in the market."
Machines and Humans, Hand-in-Hand
For the future of AI and its relationship with humans, Schmidt believes AI would not cause large-scale unemployment and risks.
In his opinion, AI may develop by combining human and artificial intelligence in the future, so that machines are smarter and can become good helpers of people. For instance, a chemist may be unable to remember all papers, but can use shorter time and improve scientific research ability by dialoging with and asking questions to a corresponding machine model.
Schmidt took "dependency ratio" as an example when talking about unemployment. The aging population and longer lifetime of people make everyone's burden of "supporting the elder" heavier. But the sustainable development of economy is energized by AI.
When it comes to AI competition, Schmidt talked about current competition in the AI speech area. In his view, both Amazon Echo and other companies use AI to set up a universal language in the most suitable way to them. So, the current competition in man-machine interaction and machine-machine interaction is fierce and such competition surely takes place in China, too.
As for the worries about AI possibly being controlled by a minority elite and giant companies and causing social unfairness and monopoly, Schmidt believes no such thing would happen. He thinks AI will benefit everyone like cell phone does, and many venture companies will grow into new giants thanks to AI. "China and America have more entrepreneurs. They can iterate continuously. This is a time of intelligence, and more companies are about to come into being, like companies operating in the areas of unmanned vehicles, medical treatment, smart traffic and distributed smart cloud. When we apply TensorFlow to various industries, we can keep upgrading and making innovations," he said.