What is the definition of AI? I always wanted to ask an AI researcher this question. My chance finally came when I interviewed Mr. Kenichi Suzuki, the director of GAiMERi, an AI Research Division of GLOBIS.
Introducing AI: More than Machine Learning?
“AI basically refers to an algorithm,” Mr. Suzuki said. According to him, AI can be interpreted as a part of software. But, he added, “when the business media uses it, they usually refer to machine learning in general.”
The simplicity of the definition struck me. Algorithms were originally applied in technology development by computer scientists trying to achieve pattern recognition, so that when similar problems arose, the scientists were better able to solve them. This technology has grown into AI, which can do just that.
A Story of Scope and Speed
The discipline of AI has evolved in fascinating ways over the past decade. As Mr. Suzuki pointed out, although machine learning is what most people think of as AI, the concept is way bigger than this. But perhaps the most impressive thing is not AI’s scope, but its speed. Neural networks, speech processing, computer vision, robotics—these and many other disciplines have seen explosive growth in the past decade
If we were to draw parallels between AI and information technology, we’d see that it took decades longer for IT to go from its infancy to its business applications. With AI, multiple developments seem to be happening simultaneously. Was there a reason? A difference?
The answer was open innovation. As Mr. Suzuki explained, “With open innovation, most things are shared. For example, coding and academic papers. In the past, if I wanted to access academic papers, I had to access them through a society, which usually sold the paper. So it was not so easy to get the latest information.”
He went on, “Now, as soon as papers are written and published, they are posted on websites.
Anyone can get access to the latest papers. And within several days or weeks, someone will kind of code them and share the code on the internet.
In the past, there were so many obstacles and limitations to access knowledge. But now, everything is open. That has helped a lot to accelerate innovation in the field.”
The power of sharing and collaboration has really played a key part in the evolution of AI technology. Open innovation has helped many other technologies, as well, such as blockchain and IoT. As much as that has helped with innovation, should we be concerned about having such powerful technology growing at such an explosive rate?
The Possibilities and Limitations of the Non-Human
“My concern,” Mr. Suzuki said, “is that people’s expectations are too high. AI cannot live up to them, and I’m afraid that people will be disappointed. Of course, AI will be applied to many fields, so I’m sure you will see many applications in the coming years.”
Where AI will fail, it seems, is in intellectual capabilities. While the applications of AI are limitless, there does exist an inherent cap on the technology.
Thinking about the form this cap might take, the first thing that flashed in my to mind was consciousness or sentience—the state of being self-aware. I asked Mr. Suzuki if he thought AI would ever overcome this limitation and become self-aware, as we define the term.
“I am not sure,” he admitted. “There are so many things we do not know about consciousness itself. AI might or it might not achieve it.”
His curiosity was clearly piqued, however. “It would be exciting, but I don’t think we have that kind of technology at hand. There is still a long way to go. Of course, in the coming years, there might be some genius who brings artificial intelligence to life.”
Essentially, the possibility can’t be ruled out. Unlike other technologies, AI’s potential is not confined to just assist us.
Having been exposed to various technologies, and having majored in physics, I have seen a wide array of concepts in nature that humans have mastered through ingenuity, creating value for society. However, out of all of our technologies, only AI has the possibility of becoming more than a tool, perhaps even becoming equal to humanity. Machines have already replaced a variety of low-value jobs, but with the progress of machine learning concepts such as GAN and creative AI, it seems it’s somehow becoming more difficult to predict the form that AI will take alongside humanity.