While Elon Musk, Bill Gates and Stephen Hawking have shown great concern over the development of artificial intelligence, Andrew Ng, the Chief Scientist at Baidu, told the media in Beijing on Monday that “worrying about AI today is like worrying about overpopulation on Mars”.
Ng took the position of Chief Scientist at Baidu in May 2014 and has overseen Baidu’s entire research branch ever since. Baidu now has three major research centers including the Silicon Valley AI Lab (SVAIL) in California and the Institute of Deep Learning (IDL) and Big Data Lab (BDL) in Beijing.
As the founding member of the Google Brain project, Ng was called “the man behind the Google Brain” by tech news site Wired. Now he’s the man behind the brain of Baidu. We translated part of his interview with some Chinese media on Monday. From his answers, we can peek at Baidu’s ambitions for edgy tech projects and see what the future of Baidu will be like.
About Artificial Intelligence
Q: How will artificial intelligence develop next?
A: To be honest, I feel it’s very hard to say. Actually, artificial intelligence (AI) isn’t one thing but a methodology. We use AI for autonomous driving, voice recognition, search and anti-spam email filters to make our data center more reliable and dependable, which is in fact a lot of things. We’re using AI to improve a lot of services such as Baidu’s food delivery services and our new financial service. Also, we can embed AI into the IoT (Note: Internet of Things) and all those appliances. Autonomous driving is a very significant part of it all.
Q: What is it like to sit in Baidu’s autonomous car, and what will Baidu do next?
A: You’ll be very excited at first but after five minutes you’ll be bored because you have nothing to do.
Autonomous driving will be put into use in public transportation first. We’re looking at initial routes of about 20 miles, like a big circle or a route that goes back and forth. If governments all over the world collaborate with tech companies, we’ll be able to hit the road safely in a few years.
Q: Elon Musk and Stephen Hawking both ask us to be cautious when it comes to AI, what do you think about this warning?
A: I said a year ago that “worrying about AI today is like worrying about overpopulation on Mars”. Maybe several hundred years from now there will be people on Mars, but now, I don’t know. I think most people are not so clear about what AI can do and what it can’t.
About Deep Learning
Q: How is Baidu’s deep learning research progressing?
A: An increasing number of products at Baidu are now integrated with a neural network, primarily because of Baidu’s internal deep learning platform named Paddle. Because of this platform, all engineers in the company can make use of deep learning smoothly.
We find that many engineers have begun to apply neural networks in their daily work, including some products we never think of in AI. For example, now we’re using it in anti-virus functions, search and also advertising. At our data center we also use a neural network to detect whether the hard drive will malfunction. There are quite a lot of products we would never think of without the platform. It helps our engineers be innovative.
Q: Is voice recognition ready to be applied on a large scale yet?
A: Baidu’s voice recognition technology improved a lot last year, but it hasn’t achieved the effect we want yet. For one thing, the computer can only be more accurate than a human when it’s processing a very short voice message. For longer ones, it’s not as good. Also, at this stage, voice recognition can only note down what you said. If we want it to know what you mean, we still have a lot of research to do.
About Online-to-offline (O2O) business
Q: What do you think of O2O business? Why doesn’t the U.S. have mature O2O companies?
A: O2O isn’t as developed in the U.S. as it is in China. If you tell Americans that you can just tap on your smartphone and there will be people that come over to wash your car, it’ll be very hard for them to understand because they don’t have the experience.
In China, it’s very common. But in many other countries, they’ve never seen things like this before.
Q: Why so?
A: I think one of the reasons is that people are more used to smartphones in China.
In the U.S., most people use laptops or desktops before they use a smartphone, so they’re used to that. It’s totally different using PCs compared to smartphones. Smartphones are always by your side so it’s much easier to use them to do a lot of things.
In countries where people are more used to PCs, we need to develop the habit of using a smartphone first. Another important factor is that China has a lot of big cities with high population density, which means O2O is less expensive.
In fact, I think the mobile ecosystem in China is more competent than in the U.S.
What do you think of Ng’s answers? Do you think Baidu can be a strong competitor in the field of AI? Tell us!
(Featured image from NetEase)
