Watch the video below to discover the future of human-in-the-loop programming and how Quickcode is revolutionizing HS compliance with the perfect blend of human intelligence and machine learning.
0:36 Fully Human Approach
1:18 Fully Automatic Approach
Well, thanks for having me today. My name is Gary King. I’m the Weatherhead University Professor at Harvard University and director of its Institute for Quantitative Social Science, and most importantly, I’m co-founder of Quickcode, which I’m absolutely enthusiastic about.
You asked me to talk today about a human-in-the-loop programming. The way that I would think about it, as you can imagine, a continuum from fully human to fully automatic, okay?
0:36 Fully Human Approach to HS Classification
So let’s consider the two ends of the continuum. Fully human, you know. Let’s imagine you wanted to do HS classification yourself. You could get it right if you put in enough effort, you could get it right, but you probably don’t want to spend all day every day doing this. You could hire somebody to do it, but you probably would like them to have some help, and how much do you really want to rely on just one person to do it? It doesn’t scale no matter how good they are, it doesn’t scale. So you could hire a whole bunch of people, but as I said, it doesn’t scale, so that doesn’t work all by itself, even though it might be the right thing, and it probably is the right thing for one code, it isn’t for a whole lot of them.
1:18 Fully Automatic Approach to HS Classification
So now let’s think all the way at the other end of the continuum, where it’s all completely fully automatic. What is fully automatic mean? It means like having a driverless car where you don’t tell it where to go. It’s not very useful; or a completely quantitative solution is like an Excel spreadsheet with no row and column labels; it’s completely useless. So you probably don’t want to be at the fully human end, and you probably don’t want to be at the fully automated end.
A hundred years ago, if I told you that I had a machine to add up numbers faster than you could, you would have thought I was crazy, and of course, it would have been 100 years ago. Now, of course, you would never think that you would, that you should be the one adding up a bunch of numbers. You would just take the results from the computer. But the computer adding up a bunch of numbers where you don’t tell it what to add up is going to produce nonsense.
So the thing that always works is the human-in-the-loop. The thing that always works the best is the intelligence from the human and the raw power from the machine. That’s what we need. We don’t replace the human. We amplify human intelligence. We augment the human. When you write, you might use Google Docs or Microsoft Word or something. It doesn’t write for you, even though it could help you. It can help you write, but what it does is it helps you. It might suggest things, but you’re not going to take it without looking at it.
That’s what Quickcode does. What Quickcode does is it takes the intelligence from a person, even how much time and bandwidth and attention they have and expertise they happen to have in this particular area, it takes all of that, and it amplifies it and expands it in ways that no human could possibly do because you can’t put thousands of codes in your head. It’s just that nobody’s head really can fit all of that, and so we can amplify human intelligence and produce a better answer than a human could by themselves and better than a fully automated system could by itself.
Human-In-The-Loop HS classification
So at Quickcode, we do human-in-the-loop HS classification. We’d like you to be that human, and we think we can improve your HS classification. You should have a look at Quickcode.ai, and you can see what we do. If you want to know about me, I’m at garyking.org, and we hope to see you at Quickcode.
Thanks very much.