If you don’t yet know who Andrew Ng is, he is the Founder and Lead of Google Brain (Deep Learning) project, Adjunct Professor of Computer Science at Stanford University, Co-Founder of Coursera and author of one of the best (if not the best) Machine Learning courses.
In other words he is to AI, what Justin Bieber is to teenage girls.
Few days ago Andrew Ng twitted video highlights from WSJDLive Conference, where he spoke about Artificial Intelligence (AI).
Here is the original twitt:
— Andrew Ng (@AndrewYNg) June 14, 2017
I found it so insightful, that I wanted to transcribe every word, to help me internalize what he said.
Here it is:
Whatever industry you work in, AI will probably transform it. Just as may be about 100 years ago Electricity transformed industry after industry. Everything from transportation to communications, manufacturing, healthcare was transformed with devise of electricity. I think that today we see a surprisingly clear path for AI to transform all of these industries. So I actually hope that whatever industry you are in, you’ll figure out how to leverage AI, because I think it will create new winners and losers in almost every category.
One very common pattern you see in a lot of industries is that first industry is digitized, so that activity is moved to computers, so that creates data. Thats the first kind of IT transformation wave. And then with the data, that give AI an opportunity to come in and eat the data, and automate decisions or do things more intelligently.
So for example online advertising realm. Possibly the single most lucrative application of the AI today may be online advertising, right, deciding what ad to show people. Because the online advertising realm has always been a digital realm, so there is tons of data and the AI for that is very sophisticated today.
Healthcare is a little bit further behind. In the United States over the last eight years or so, the Affordable Care Act or the devise of electronic medical records is creating data and now it is at the phase for the industry, for AI to come in and eat that data.
I think education is bit further behind. A lot of education is still offline, analog.
Then a panel member asked him a question:
Q. So let’s be predictive, you mentioned my 9 year old twins earlier. So they are in the phase now, when they are coming to me and saying what they want to be when they grow up.
What things that they might say to me, should I tell them sorry that won’t exist when you are an adult. Radiology is the one that you mentioned before that’s in the crosshairs, right? What else is in the target zone?
Andrew responded with:
A. OK, the Radiology, fine. If any of you have friends or children or whatever studying in a med school, AI is getting much better at reading radiology images frankly. So if any of your friends are going through medical school and graduation with a degree in Radiology, I think they’ll have a perfectly fine 5 year career as a Radiologist.
The broader pattern is that in any task where there are a lot of people doing relatively routine repetitive work, when a lot of people are doing very similar things, that creates a very strong incentive for AI teams to come in and automate that task.
So … self driving cars, I think that will displace a lot of workers. Call center operators, many people doing relatively similar things, it will displace that.
So there is actually one other rule of thumb… Just think, what can AI do, it’s a bit of mysterious concept. So there is one highly imperfect rule of thumb, that I sometimes often like to use, which is almost anything that a typical human can do with less than 1 second of mental thought, we can probably now or in the near future automate using AI. Such as a security guard looking at a video feed, saying are the people in this, are they doing something suspicious. That tasks is actually taking a lot of one second judgment things and stringing them together, and so I think that a lot of that job can also be automated.
I’d like to conclude with another quote from Adnrew:
If you’re trying to understand AI’s near-term impact, don’t think “sentience.” Instead think “automation on steroids.”
— Andrew Ng (@AndrewYNg) May 1, 2017