Executive Chairman, KNIGHT-SWIFT TRPTN HLDGS INC
Dialogue | Kevin Knight and Dr. Sebastian Thrun - Global AI Summit 2022
About Kevin Knight
Kevin Knight, executive chairman of Knight-Swift Transportation Holdings, has spoken about the company's operations and the trucking industry in several interviews and company messages. In 2022, he discussed the merger of Knight and Swift, describing it as a "match made in heaven" and stating that the combined company would be the largest truckload carrier in North America. He has emphasized the importance of trust between the two companies and has said that his goal is to improve Swift's financial strength while providing growth opportunities for Knight. Knight has also expressed his view that autonomous trucks will not replace drivers in the foreseeable future, stating he will "always fight for our driving associates to be in the cab" and envisioning drivers' roles evolving to be "much like a pilot." Knight has frequently praised professional truck drivers, calling them "the hardest working, most patriotic group of individuals" and stating that the economy depends on their work. He has expressed gratitude for drivers' sacrifices and said the company will never take them for granted. In a 2022 appearance at the Global AI Summit, Knight discussed artificial intelligence, defining it as tasks where machines are approaching human-level abilities and stating that he does not see artificial general intelligence coming soon. Separately, in a 2021 interview about a personal Airbnb business, Knight described building a portfolio of 12 homes through referrals and low customer acquisition costs.
Transcript (14 segments)
K
Kevin Knight0:00
Great to be here. All right, very good to be in Saudi Arabia. Thanks very much for the invitation from the organizers. We're going to talk about what is AI and what isn't. Is AI taking away everyone's jobs? And is AI going to take over humanity? So just a couple of topics. I'll start. What is AI? What we talked about for a long time was AI is a group of tasks where those tasks are ones where people currently do better than machines. For example, it used to be playing chess was an AI problem because people could play chess better than machines. Recognizing speech, being able to translate language, driving a car—all those things were AI problems. But now machines are getting better at all those things and approaching human level abilities. So it seems like it's time for a new definition of AI. So maybe Sebastian, you could say something about what you think AI is and isn't.
S
Sebastian Thrun1:03
First time, extremely delighted to be here with Kevin and be here in the Kingdom. I think the incredible enthusiasm for AI really shows, and it's a great thing. You're so enthusiastic because I think this technology is going to change everyone's life. We've been asked to say what this is and isn't. It's very simply to define: you raise an AI the same way you raise your children. You can teach a computer now by giving it examples, and as your children become more and more intelligent, so does your artificial intelligence. Eventually, it'll be able to do pretty much anything that people do. Anything repetitive, anything that we do over and over again, AI is doing about as good as people can do, and wait a few years, it'll do better. So that's my best way to explain AI: it's a piece of computer software that will eventually become really, really smart.
K
Kevin Knight2:00
Yeah, and I think our machines become more by learning from data and learning from instructions that people give. So let me give an example. We worked for a long time on translating between human languages, like translating Arabic to English. When we started doing it in this way of teaching the machine and having the machine learn from data, at the beginning—this was not that long ago—it was a big disaster. We translated one page per day, very slow translation, and the quality was not very good. You look at the output, you say, 'I know what it's about, but I can't really get all the details.' And now, within a very short amount of time, using AI technology, we're able to make much more accurate translations. I'm sure you use it every day. In fact, it's really amazing to me, the pace is so accelerated. You can also use it for free on the web. Even now, I think this summer, if you use Zoom, you'll be able to start to talk to people who speak other languages and have what they're saying to you captioned in your own language on the screen. So this is just one example. The pace is just super, super accelerated.
S
Sebastian Thrun3:15
Yeah, I thought of the Google software project, the self-driving car project that became Waymo. Again, their vision was you can just teach computers how to drive. In the beginning, it was just like what Kevin said: every two or three minutes, a safety driver had to grab the steering wheel to keep this stupid machine from basically creating a car accident. And now the Waymo cars have driven 30 million miles without any accidents. It's kind of amazing. It's an example where machine skills at this point have outflanked people's skills. In the future, you shall be so happy to have a machine drive you because it's safer than people driving you. I got my driver's license when I was 15 years old. I think that looking back, that seems like crazy. It looks fantastic to have this automatic driving.
K
Kevin Knight4:10
So, are these machines, if they're getting good at all these tasks, are they going to take over everybody's jobs? No, no. What's always happened in society—and that's really important, always happens—is we make machines to make us people stronger. We team up with machines to help us where we need help. The same is true for cars. Cars have not replaced our legs, but together with a car, a human is stronger and able to move about faster. The same happened in agriculture when we brought steam engines and so on to farming. So it's always been a situation where we build machines to make us stronger, but then we people are resilient. We find new and better jobs, become more creative and more engaged, and bring society forward faster and faster. So don't think of AI as replacing human labor; think of it as augmenting human labor and guiding human labor. I think the Kingdom has something on the forefront of doing this. I remember I met His Royal Highness, the Crown Prince Mohammed bin Salman, about five years ago, and we talked about this. We talked about what's going to happen to the Kingdom and to the Middle East. He then immediately recruited my company, Udacity, to train—I think we've now trained over 65,000 AI experts in this country alone, which is completely amazing. I think it takes some vision, some visionary capabilities, to understand how to lead a country into the future. I think the Kingdom is standing out at the forefront of the Middle East and the world doing this right now. So I'm very, very happy to be here.
I agree. So throughout history, mechanization has replaced some jobs, but those have always been created by other jobs. There are so many more jobs on the planet today than there were just a few decades ago, despite all this. But I think some people think AI is different. AI is the first technology out of all these technologies that really will put everybody out of work. I kind of disagree with that. If you take automatic translation, the example I just gave you, you'd think that would have by now put all the human translators out of work, but it's not the case. There are more human translators now than ever, and they're doing very important things—translating medical documents, legal documents where we can't rely on a machine to do it wrong, make mistakes. The reason there are so many human translators is the machines aren't replacing them. The machines are translating stuff that wouldn't be translated otherwise. So machines are doing these jobs, but they're not replacing a human taking a human from doing what they were doing and then doing that job instead. They're doing all these new jobs. Every time you use Google Translate or Microsoft Translator, you're not taking away a job from a person. No one would have translated that for you anyway. That opens up this huge economic potential that we're all talking about here. All these new things become possible with the technology.
S
Sebastian Thrun7:15
Yeah, and I'm so excited that we are able to be here together. You're the leaders of the Arab world, and it's great you're taking time out to think about technology, specifically artificial intelligence. Know that there are a lot of people who want to scare you and say it's going to enslave us. Don't listen to those people. It's an amazing tool. Like any business thing you do, you can bring AI and make it stronger and better and more resilient. I think that's why you're here: to learn from us, from our experiences, and then bring this back into your governments, into your companies, and to your businesses, and see how you can make your own business stronger. It's going to be a big change because now we have technology that can do any amount of repetitive work that people tend to do—be it an accountant, be it a pilot, a driver, a certain type of lawyer, speaking a medical doctor. All these fields will be infiltrated by AI to make people stronger. I'm so glad that I can be part of your journey here in the Middle East, help Middle Eastern people to learn and understand artificial intelligence.
K
Kevin Knight8:25
I think it's good that we talk about how machines not necessarily replacing people—sometimes they do, sometimes they don't—but most often the machine winds up working together with the person. In lots of tasks, the machine can help correct errors made by the people, and people can correct errors made by the machines, or they can work together to solve the problem. I also don't fear spending a lot of time fearing AI in the sense of killer robots from science fiction. If I was going to spend time fearing things and fixing things, it would be more about how people might misuse AI, either inadvertently or on purpose. People are very creative at coming up with ways to use technology in the wrong ways, so that's something to worry about.
S
Sebastian Thrun9:22
Yeah, I think the biggest thing I would worry about is to get ready. How can you get ready for something that's global, that's going to involve all jobs, and it's technology that's really novel? I've heard this several times this morning: how important training, skills training, education has become in the region. It's not just Saudi Arabia. We've been working with Dubai, training more than one billion—error, quotas—with Egypt, where we're responsible for a huge number of new freelancers. I think that mindset of saying, 'Let's get ready, let's get ready as a country, as a culture, as a people,' is really important. If you do that well, then you'll be part of the leadership leading the world. If you do it poorly, you're going to be swept over. I don't wish you that. So I would say the most important thing is be optimistic and get ready.
K
Kevin Knight10:19
Yeah, I think people will always, no matter how good, people are still going to complain. If you think about air travel, I just took a long trip over here, and when you listen to people talk about air travel, it's nothing but complaining. But obviously air travel has been a huge economic boom for mankind and allows things unimaginable 100 years ago. It'll be the same with AI. We should make it as good as we can. It's going to revolutionize everything, but be ready for a little bit of complaining. So I don't know if this session helped you. I mean, the question 'What is AI, what isn't?'—it's sometimes hard to define because it's such a big field with such big names. But what you could take away from it is that the technology is actually quite basic and quite simple. It is something that will affect us as people. It can memorize things, it can recognize things, it can make decisions with us and for us in a way that's never seen before. In many ways, it's exceeding what people can do. Today, the best chess players are AI, the best Go players are AI, and the best car drivers are AI. But it's our job as people to redefine ourselves and make this technology our very, very tight partner. So as we go forward, we can really utilize it and make the world a better place.
S
Sebastian Thrun11:40
Yeah, a lot of the excitement is coming from this new technology. Even just maybe three years ago, people used to say AI systems are very narrow—they can only do one very narrow task, and that's what differentiates them from humans. But within the last three years, we've got a lot of systems now that can transfer their tasks. So if you teach a computer how to translate from French to English, it used to be when you wanted to teach it to translate Spanish to English, you had to start over from scratch. But now the machine will say, 'Hey, wait a minute, I already learned how to translate from French to English. Spanish is very similar.' So with just a few examples, it can learn that task. So transferring knowledge between different tasks, and also self-study—this new technology, everybody knows about AlphaGo, that can study by itself to learn to play chess and Go programs. But this is now happening in natural language processing, where systems go off and study by themselves, and they come back kind of knowing English. So if you want to teach it to do something like answer questions or categorize movie reviews, the fact that the machine kind of knows English already makes it very easy to do a wide range of tasks, not just one narrow task.
K
Kevin Knight12:53
This is really interesting to me. I studied AI even when I was an undergrad, like 20, 25 years ago. In fact, we knew each other back then. I wrote a master's thesis on AI and machine learning, and a PhD thesis in 1995. But back in the day, it didn't really work. It was like nice academic—okay, you get a thesis, but no world impact. We had what's called the AI winter back in the day, which was lots of promises, very little results. And last—Kevin is fine—last three to five years has completely changed. We now take the same ideas that are sometimes 30 years old—you can see the next speaker talk about this—and put them into very fast computers with very big data sets, and all of a sudden you get this magical thing where really interesting stuff is happening. Your AI system can augment or replace the best human doctors or the best human lawyers today, or of course play chess better than anybody else, or drive cars. That's really an inflection point. It's an inflection point, and that's unique in the history of humanity. It's a great opportunity. If you don't fear it, if you embrace it, and if your staff, your people embrace it, you can be at the forefront of these developments and harvest them for the betterment of society.
S
Sebastian Thrun14:09
And since there's just a couple seconds left, how about artificial general intelligence? Now we're back to science fiction: robots that walk around among us and have human-level or superhuman-level intelligence. Is it coming or not? I don't see that coming yet, and I don't think we want to build machines that are smarter than us. I think we build machines that are good for us. It won't be the machines enslaving us. Look, there's a lot of talk about it, but all the progress today seems to be specialized—really in game playing, or language translation, machine translation, or in self-driving, and so on. What's your thought?
K
Kevin Knight14:42
Might be coming, might not. Who knows? But people often say, 'Would you rather have a robot doctor that's 90% accurate in its diagnosis, or a human doctor that's 80% accurate?' Different people have different preferences on that. But I think what you really want is a human doctor working with a robot assistant, and then you get the best of both worlds. Well, it's been a pleasure being here with you. Thank you so much for inviting us. I hope that our conversation was somewhat enlightening to you. Thanks to the organizers.