Andrew Ng, who co-founded Coursera, driven by desire to improve world
US-based scientist born in Britain to Hong Kong parents helped transform online learning
Students wondering where a life in scientific research might lead them should ponder the career of eminent scientist Andrew Ng.
The 39-year-old Ng, who was born in Britain to Hong Kong parents, works simultaneously in the corporate sector and academia. He's chief scientist at Baidu Research, an institution owned by the search engine giant that's based in Silicon Valley, California. At Baidu, Ng is conducting research into computer science subjects such as image and speech recognition. Ng is also associate professor at the department of computer science and the department of electrical engineering at Stanford University, California, a private research institution.
Ng teaches machine learning at Stanford, a subject which aims to develop computers that simulate human thought processes. But he's probably best known for co-founding the popular massive online open course (MOOC) Coursera in 2012, a for-profit project that was the result of his after-hours research into online education at Stanford.
There are differences in conducting research in the corporate sector and in an educational institution, but the two experiences are more similar than is commonly thought, he says. Both can offer a high degree of freedom, although university research can be heavily constrained by the availability of funding. Those conducting research should take a good look at what they are trying to achieve, and then weigh up which approach will best facilitate the achievement of their goal, Ng says.
"I am always mission driven, and I always ask myself what I want to be working on, what project excites me the most. I figure that out, and then find the best place to do that work.
"The mission that excited me a few years ago was universal access to education. I thought that starting a new company was the best way to execute that mission, so I formed Coursera. But after this, I wanted to return to the artificial intelligence mission and develop new technologies that will have a lot of impact on people. I decided that Baidu's research environment would be the best way to execute that mission. No university is the best place to execute every mission, and no one company the best place to execute every mission."
As a computer scientist, Ng says he wants to transition the results of his research into products and get them out into the world as soon as they are ready, so that they can benefit end users. Universities are seldom good at making this transition, he says, which is why it's a good idea to work in the private sector sometimes.
"You can often get a good amount of resources to get your research done in a university," he says. "But most universities are not product organisations. So it's difficult for universities to affect that many people directly, although they can indirectly affect huge numbers of people by publishing papers of research and hoping that some third party will take them up and carry the ball across the line.
"There are so many problems in the world worth working on, and so many discoveries to make, you have to make a choice. My preference is to focus my efforts on solving problems that will help people. I wrote a lot of research papers when I was younger, and some of these captured the imagination of the research community. But as I got older, I worked out how to transition these research papers into helping people, and that's what excites me today."
Coursera is one of his ideas that transitioned in a big way - more than 13 million users have signed up for the MOOC since it started in 2012. Ng says it grew out of a personal interest in the possibility of free universal education, and didn't involve much of his research in machine intelligence.
Indeed, the scientist felt that it was so far removed from his day job that he initially worked on the idea at home in the evenings and at weekends. It was only when the project started to come together that he brought it further inside Stanford. "I had a strong interest in free online education, and I was interested in what videos and formats would work for it. A lot of education workers were very sceptical about what computer scientists were doing. It was only after the first visible success of MOOCs that they started to take it seriously."
As with all educational methods, the most important thing about a MOOC is the relationship between the teacher and the student, says Ng. "A MOOC teacher needs to focus on doing what's best for the learner. The most successful MOOCs have been designed by instructors who have deep empathy for the people who are taking the course, whether it's about machine learning, finance, history, art or any other topic."
That sentiment is echoed by Dr Beth Simon, Coursera's principal learning specialist on the course success team.
"The most important place to start when designing a MOOC is to clearly identify the learners you want to reach," Simon says. "What kind of background do they have? What are they hoping to change in their life? How does that match to specific knowledge and skills you want to help them develop? In essence, how will learners be different at the end of your course? With this foundation, instructors are much better prepared to define the kinds of tasks and assessments with which they want learners to engage," Simon says.
Ng, who has also conducted research into robotics, worked for Google, developing the Google Brain project in 2011. He joined Baidu a little more than a year ago as the company's chief scientist. He's responsible for building a new research organisation that will create next-generation technology in areas such as speech recognition and image recognition. "In the California lab, we say that our mission is to develop artificial intelligence that affects hundreds of millions of people. I've spent a good part of my life trying to develop artificial intelligence technologies, and I've been fortunate to have a few successes in getting technology into the hands of users."
Science students should be driven by a desire to improve our world, he says, and computer scientists have the power to do that. "Our goal is to make the world a better place, and leading scientists should take the responsibility to do whatever it takes to achieve that," he explains.