Andrew Ng

I think the first wave of deep learning progress was mainly big companies with a ton of data training very large neural networks, right? So if you want to build a speech recognition system, train it on 100,000 hours of data. ”

Education is not about thinning the herd. Education is about helping every student succeed. ”

I find it a very encouraging sign for a society if employers are bringing online education to their companies, helping employees gain more knowledge. ”

With human inspectors, it's difficult to get even the same person to make consistent judgments. ”

Our educational system globally has not been historically great in reskilling for newer job roles. We need a new social contract to do that. For India, lack of an incumbent structure might be an advantage, where it can use digital education to leapfrog. ”

Baidu and Google are great companies, but there are a lot of things you can do outside them. Just as electricity and the Internet transformed the world, I think the rise of modern A.I. technology will create a lot of opportunities both for new startups and for incumbent companies to transform. ”

Text input is certainly useful, but images and speech are a much more natural way for humans to express their queries. Infants learn to see and speak well before they learn to type. The same is true of human evolution - we've had spoken language for a long time compared to written language, which is a relatively recent development. ”

The Chinese market is very different. One of the things that I believe is that the biggest, hottest tech trend in China right now is O2O, or online-to-offline. ”

We think that many companies view Coursera as a quality, convenient, inexpensive way to continue employee development. Is there a contract with a company that might make sense? I don't have an answer to that yet. ”

Want to train a machine translation system? Train it on a gazillion pairs of sentences of parallel corpora, and that creates a lot of breakthrough results. Increasingly, I'm seeing results on small data where you want to try to take in results even if you have 1,000 images. ”