We talk about artificial intelligence and intelligent machines as if killer robots loom just around the corner. But if we all learned the basics of AI programming, would we be less scared of it?
AI pioneer Andrew Ng is the co-founder of Coursera, an online education company that is offering classes in deep learning, a type of machine learning inspired by the brain. Marketplace Tech host Molly Wood talked to Ng about what this means and why he believes basic AI education should become a standard. Below is an edited excerpt of their conversation.
Molly Wood: I know that this is a growing area of importance and I wonder is it analogous to predictions that there will be people who know how to code and don’t know how to code? Like, in the future will we be split into two groups: the ones who know how to create smart machines and the ones who don’t?
Andrew Ng: Once upon a time I think we used to have debates about whether all of society needs to be literate. And today we’ve figured out that it’s actually good if everyone knows how to read and write. In the future I think that it would be good if almost everyone knows how to code. Literacy is a great way for people to communicate. And computer literacy or learning the basis of coding is a good way for humans and machines to communicate. And I would love a future where even the proprietors of a mom and pop store can write a little bit of code so that they can customize the display in their shop window. I think we have a long way to go in society to teach a lot more people to code and I don’t see a shortage of work.
Wood: We now think of reading as pretty easy to learn, that the vast majority of people should be able to learn by age five or six. Is coding similar in terms of difficulty?
Ng: I believe, and science seems to support this, that almost anyone can learn almost anything. And certainly I think a lot of people, almost everyone, can learn to code and learn to build AI systems.
Wood: What kind of prerequisites are there to learn about deep learning? I assume I should not sign up for this course.
Ng: To jump into the deep learning AI course you need basic programming knowledge and a little bit of prior experience with machine learning. I would hope that even if you don’t have that and just know how to write simple code that you might be able to follow along with the course.
Wood: I think there’s been a lot of hype around AI, it’s almost a buzzword. Like you hear companies essential say, “I put some AI in it and now it’s way better!” But it’s hard I think to sometimes understand what that really means. Do you have some examples of what systems, what products artificial intelligence might actually change?
|"Whole professions could wink out of existence," a conversation on the future of work|
|The most robot-proof job of them all|
|More from our Robot-Proof Job series|
Ng: One funny thing about AI is that once it works people seem to stop thinking about it as AI. A part of me thinks that when we actually have self-driving cars – these AI self-driving cars that seem so sexy and exciting today – I hope that when we reach that day you just think of it as your car. It won’t be an AI-powered car anymore.
Wood: And you’ve also said, on the topic of drivers and many others, that Silicon Valley needs to recognize that this technology is going to take a lot of jobs away. What sort of responsibility do you feel in that area?
Ng: People don’t like this term, but I think the East and West Coast, quote, "elites," create a lot of wealth. They’re actually very good at that. But I think to be plenty honest we may have also left some of the country behind. So if we want to create not just a wealthy society but also a fairer one, I think there’s important work ahead of us. The solution I favor is conditional basic income where we have the government support people to keep learning so that anyone who loses a job has a better chance of learning what they need in order to reenter the workforce and contribute to the tax base that’s paying for all this. You know, I think there’s a lot to be said about the dignity of work. And we should put in place the support for people to learn what they need so that every man and every woman has a shot at being able to do meaningful work.
Wood: What about the education part of that? I mean I assume that you see Coursera as part of an attempt to rethink education. This model hasn’t necessarily been as disruptive or displaced higher education as much as people thought. But do you think it plays a role in that retraining?
Ng: I think Coursera can play a huge role in helping retrain displaced workers. I’m not convinced that Coursera by itself would be the entire solution. But what we’re seeing is that there are a lot of people post-college that are taking Coursera courses to keep improving their careers. The old model of education where you go to college for four years and you coast for the next 40, it just doesn’t work in today’s world where knowledge decays so rapidly and you need to keep learning new things. And I think Coursera has helped millions of people gain those skills to keep on upscaling their career.
Wood: Very recently there’s been a big uptick in people expressing a lot of concern – and that uptick includes people like Elon Musk who are saying that we need a neural interface right now so that we can communicate with machines. What do you say to that? It sounds like you have a much more measured view.
Ng: I found that, on average, the more someone knows about AI the less fearful they seem to be of it. I think that we will build an AI-positive society that will deliver much better education, healthcare, transportation. So I am a bit dismayed when there are a few individuals that seem to be trying to slow down progress in AI, when the rest of us are working hard to accelerate progress. I’ve said before I think worrying about AI evil killer robots today is like worry about overpopulation on planet Mars. You know, it could happen. Maybe the planet Mars will be overpopulated. But I think as a society we’re massively over investing today in studying the overpopulation on planet Mars problem, and putting more of those resources than actually figuring out how to land on Mars or how to advance AI would be better for society.