(Yicai Global) Oct. 11 -- Chinese internet giant Alibaba Group Holding Ltd. [NYSE:BABA] recently hosted two technology forums in the US, during which Yicai Global held an interview with Jin Rong, dean of Alibaba's Institute of Data Science & Technologies (iDST), at Carnegie Mellon University in Pennsylvania.
The discussions centered on the implications of AI for Alibaba, the e-commerce market and humanity as a whole.
Forecast, Optimization and Personalization
Yicai Global: What is the role of artificial intelligence in terms of helping Alibaba to develop e-commerce businesses?
Jin Rong: The role played by AI is multi-fold. First, the most salient characteristic of e-commerce is its ability to match sellers with buyers, [online] forecasting and [traffic] optimization. Buyers need to identify their most suitable and high-quality customers, and [AI] can help with online predictions and the optimization of traffic. Content is very important for e-commerce businesses. Sellers want to retain customers, so they create large volumes of promotional content adapted to market demand. Video content analysis is important because it generates interesting, user-engaging content, thereby maximizing the length of time users spend on sellers' sites. [These considerations] are critical in terms of optimizing online traffic, individualizing services and categorizing media marketing content.
Yicai Global: How significantly will AI contribute to e-commerce development in the future?
Jin Rong: Let me give you an example. You must have heard of Double 11 [Nov. 11, the biggest online shopping day in China, and the country's equivalent to Black Friday]. Double 11 promotions, including banner ads, used to be manually produced, but now everything from design and webpage displays to user interaction is done by machines. A big advantage of this is, take banner ads for example, in the past, a team could only work on one project at a time, but now advertising can be personalized. In other words, if I know your preferences, say, you prefer red to green, then it can make ads more effective.
Yicai Global: A while ago, an ISIS video shown on Facebook was preceded by a perfume commercial featuring Jennifer Aniston, this was quickly deleted. This is something that might happen with Big Data technology. Are you worried that something similar may happen to your company?
Jin Rong: People should always play a part in the process. We will never let computers to do everything. We have a set of very sophisticated procedures, which include a checking system in addition to the automatic generation program.
AI Creates More Opportunities
Yicai Global: AI will replace many human workers in the future. As an expert, do you think that Alibaba will cut back on new hires as a result of artificial intelligence? To what extent will AI be able to replace people on a worldwide scale?
Jin Rong: Firstly, AI is very useful and has wide applications in our everyday lives. Secondly, I think AI can be used for other things. Take computer programs for example. Software was created on punched cards in the early days. The invention of programming languages freed people from many manual tasks and created many new jobs. The same thing has happened to banner ads too. Indeed, many coder-designers have lost their jobs as a result of advances in AI technology, but it also enables people without any experience to design their own ads. With AI, all one needs to do is to input some superficial descriptions, and the system will create something for you, you can then revise the design based on it. AI turns everybody into a designer, and I think it offers enormous opportunities. On the flip side though, some designers will lose their jobs because of it.
Yicai Global: So you mean AI actually creates more demand?
Jin Rong: Yes. Besides, it pushes the industry forward, because people no longer need to perform simple tasks. Designers can focus on more innovative, more creative work. I personally think it is a good thing.
Yicai Global: So is the demand big enough to accommodate so many people?
Jin Rong: I think this involves two questions. First, where are we in the process now? Second, where is the limit? For the first question, many people feel that they can get everything they want with existing technologies, but if we look more closely, for example, some 800 million items of products are offered on Alibaba, and no one can possibly count all of them. Without external help, most people can only access a very small proportion of the total volume, meaning that most designers cannot enter the e-commerce market. After all, it is impossible to design ads for all of the 800 million products. So if we have a smart tool that knows how to sell products by showing certain items to users at certain times, it will give small designers significantly more opportunities to get involved in the e-commerce design business. We are still far from reaching the 60-pass score, at least judging by the total product volumes at Alibaba.
Yicai Global: To my unlearned eye it is already pretty advanced now.
Jin Rong: Most designers don't have access to the e-commerce market, relative to the limited number of coder-designers who lost their jobs. Small designers and businesses will be able to enter the market, if AI can reduce the volume from 800 million to a much smaller amount.
Yicai Global: But which numbers more, the new opportunities created or the jobs that disappear because of AI?
Jin Rong: It all depends on how you look at it.
Advances in Technology Will Bring More Jobs
Yicai Global: Autonomous driving will make many drivers lose their jobs, and even doctors and lawyers may be replaced in future. This has become a source of concern for students and their parents.
Jin Rong: We still have a long way to go before self-driving cars can be accepted by the majority of consumers, but new technologies stimulate suppliers and set new requirements for a vehicle's technical performance. As a result, more job opportunities will be created, so I believe advances in technology will lead to more jobs in the marketplace.
Yicai Global: You sound very optimistic.
Jin Rong: We talked a lot about the oil crisis in the past, but people always manage to find a solution for this kind of situation.
Yicai Global: Many business magnates in Silicon Valley, Elon Musk, for example, are also concerned that AI may spell disaster for humankind.
Jin Rong: In my opinion, there will be conflicts, but I am optimistic about the overall development trend.
AI and Humans Live in Parallel Worlds and Have Different Strengths
Yicai Global: What is your advice for young people, regarding what they should study so that they can find their place in the society? They used to say that math, physics and chemistry are the most employable skills in the world. More recently some people claimed that an era of arts students has arrived. So judging from the development of AI, what skills will make people more employable?
Jin Rong: Demand will not decrease [for science and engineering students]. Machines will free people from the simplest tasks, allowing them to focus on more creative things.
Yicai Global: Here comes the ultimate question -- will robots be more intelligent than humans in future?
Jin Rong: For all machine learning activities, the most important thing is Big Data. It has been theoretically proven that [machines] must go through a certain number of trials and errors and receive enough training before they can acquire certain abilities. The volume of trials and errors and training required for humans is very low. Humans are very good at abstract reasoning, but we have no idea at all of what abstract reasoning means for machines. Machines are not capable of reasoning before they make a certain number of inferences, which is totally different from how it works with humans. We still do not know how to turn reasoning into a mathematical operation. So machines will never outperform humans in this respect. We live in two parallel worlds. Computers and people have different strengths.
Yicai Global: Will we see many robots in our daily lives?
Jin Rong: First, we need to have a clear definition of what kind of robots we are talking about. Do you mean robots that can communicate with people, have an appearance similar to people, and, the most difficult part, robots that can understand natural language, or can answer questions in natural language? Watson, a computer developed by IBM, beat human contestants in the US television quiz show Jeopardy. So we feel that machines now can outplay humans in listening comprehension. But that is not the case. We use the same technology too, for example, in developing intelligent customer service systems. Actually, these robots work based on data analysis, which does not involve any comprehension skills. The same is true for Watson.
Yicai Global: So could you give us an example of something robots cannot do that involves comprehension?
Jin Rong: That would be something that requires reasoning. There was a quiz show in the US. At first, all the questions asked were about facts, and they were easy [for robots]. Later on, the host started to ask questions that require reasoning such as "Who was the second non-American person to walk on the moon?" Actually, it does not really require reasoning, but to answer the question, you must understand what it is about. This is very difficult for machines. Machine learning is based on data analysis and statistics. Comprehension and reasoning are simply beyond that capacity.