David Gentle, director, foresight and planning, portfolio strategy, marketing unit, Fujitsu Ltd, talks about the evolution of AI and the importance of cyber security.
From using artificial intelligence (AI) inside retail stores to helping the deaf catch the characteristics of sounds, to running a factory in China, as AI matures, digital companies like Japan’s Fujitsu are further developing it to a level wherein it becomes an integral part of consumers’ life. In a conversation with Anushree Bhattacharyya, David Gentle, director, foresight and planning, portfolio strategy, marketing unit, Fujitsu Ltd, talks about the evolution of AI and the importance of cyber security. Gentle leads the creation of Fujitsu’s corporate vision on behalf of the company’s international business. Excerpts:
While AI has been in existence for some time now, what makes its role more important than before?
AI probably is the big buzzword for the moment. While the older generation techies would say that there is nothing new about it as it was first developed in the 1950s and ’60s, in the last 10 years a combination of things have happened, for it to play a bigger and more important role. With new technology, one now has access to faster computing power, besides access to more data than before. So, instead of programming a machine to learn all kinds of eventuality through pattern or picture recognition, there are other means to do the same job. For example, there are two ways to make a computer learn how to cross a road. One can either put in all the instructions through programming or give it millions of hours of YouTube footage of people crossing roads. It is then provided feedback on which are the right and wrong ways of crossing a road. The machine then learns the principle through a new mechanism called deep learning. Deep learning as a new technology is hugely powerful and has transformed AI.
What are the limitations of AI?
We worked on a project in Tokyo to build an AI which is able to clear the entrance exam at University of Tokyo. There are two parts to the exam—one is a math exam and another is a language exam. While the AI cleared all levels of the maths exam as it largely involved usage of numbers, algorithms, etc., in the language test, the system was only able to clear a certain level. It wasn’t able to get to the higher level because one has to understand the context to be able to answer language related questions and it is not much about pattern recognition. So there is still a long way to go for AI.
With big platform companies tightly guarding their respective cloud businesses, will we ever see any kind of collaboration in cloud computing?
Yes, we will see a lot more collaboration between platforms, but a lot will depend on the business model. For example, large platforms such as Facebook follow a business model where the aim is to try to be the biggest platform, so perhaps it will not find any incentive to collaborate and share. At the same time, collaboration will depend on the nature of the service. For instance, at St. Carlos hospital in Madrid, we created an AI using data from about 35,000 patients with different ailments. The AI then used the data to learn the patterns related to a symptom of a disease. It then made an analysis to suggest medicine. However, such kind of platforms cannot be opened to other sources, as medical data requires a minimum level security. In the future, on one side big players will operate alone, on the other smaller firms will join hands in the business of cloud computing.
Where does cyber security stand given that AI uses user information to further recommend on topics such as shopping?
Security is very critical, particularly when we are talking about a border-less world. While a lot of people in the internet industry say privacy is dead, I don’t believe that. I think there will come a point when privacy will become a big part of certain customers’ preferences, while the rest will be less concerned about it. It also comes down to the point what the customer wants from her relationship with Internet firms/brands. For example, when she puts her personal information on Google in an effort to find an answer to a few questions, I think under those circumstances it is a fair barter. So security too will evolve with the needs of consumers.