Click here for the Chinese version.
In the last day of Global Machine Intelligence Summit 2017, a few guest speakers are invited to a discussion panel, where they exchanged their ideas in the current AI market in the world. Suji talked thoroughly about the Japanese market, where both the researchers and users show some unique features. Click for the complete video.

Hello everybody, I’m very pleased to be here. Thank you so much for inviting me. as Chain just explained, I’ve been writing about AI for a couple of years for MIT Technology Review.

It’s very clear that AI is already an international phenomenon - the founding fathers of AI come from all over the world. We witness those activities in the US, and as we’ve just heard, Canada, too. There’s an amazing amount (of activities) from many different countries. And as you’ve heard from the last couple of days, there’s a really exciting amount of stuff coming out of China. And at the same time, I think it’s very important to look at this issue of globalization and how this is going to pan out. As Chinese companies look abroad and as other companies in the US and Europe look into the Chinese market, it’s going to become very interesting in the next few years.

We have a fantastic panel for you to talk about this. I’d like to introduce the speakers individually. First, we have Wesley Mukai, who is the CTO of GE Transportation Digital Solutions. Next, we have a person you’ve just heard from, Jean-Sebastien Cournoyer, who is the co-founder of element AI and a co-founder and partner of Real Ventures. Thirdly we have Adam Kell who is also a partner of Comet Labs. Last but by no means the least we have Suji Yan who’s the CEO of Sujitech, a Japanese AI company. Thanks, guys. Thank you.

I just want to start off by talking about the question how you take an AI company international. What the challenges are? What the opportunities are? Wesley, I know this is something you have some visibility on, being working for a big international company. Tell us a bit about the challenges and the opportunities of going international and maybe going into China.

Wesley:

In GE Transportation, we had our focus on in the trains base. In transportation we had great success going abroad from a more the physical assets - building trains. We do a lot of work in China and South Africa. It has been very common and very easy to do. But as we got to the digital side of our business, i.e. software, and even now with more data science, I find that it’s much more challenging. It’s because of large amount of the data and the domain which resides in a country, and the knowledge such as whether it’s the analog space resides in countries, too. Exporting that kind of knowledge and applying technologies abroad in the software space has been a big challenge for us. We are right in the middle of that challenge right now

JS:

As you may remember what I described, Element is about building an ecosystem. We collaborate with research labs and universities, enabling their professors to work directly with large corporations. As we expand internationally, we are building these clusters in the given markets. We started working on the Japanese market. It starts with recruiting a group of fellows which are faculty members in some of the largest research labs and universities who will work for us part-time. On that platform, we will add our own research team, our applied research team, and our engineering team. But it will all come from the local people. Then, we exchange the knowledge that we’ve built in the Montreal cluster with the one here.

Will:

Is that why everybody is building a lab everywhere? For instance, global companies are putting labs in China, and Chinese companies are putting them in Europe.

JS:

You need to go where the talent is. Fortunately, here in China, you have great universities, there is a lot of great AI talent here. Japan has the same. Even Singapore has as good AI talent as well. But the challenge has been in making talent accessible and valuable for large corporations. it’s very hard to get from the hardcore innovation and then turn that into software applications that can be used by large corporations. You would rather get density in a few organizations. Then, you have every large company with you. (Smaller teams with) 5 to 15 people will struggle to build world-class applications.

Will:

Adam you are nodding your head there. I can tell you are agree with that.

Adam:

I think it’s still early days for AI. I think most of the time you know that the home country where the technology is developed is the first market that comes to mind. For the obvious reason that the founders are from that place. But I think it’s still too early to be able to tell how that kind of knowledge will be transferred. AI companies have a unique difficulty and will be datasets that they work with are generally very focused on environment. There’s a spectrum of companies that you could see being much easier to go international. When you are picking apples, for example, photos of apples look the same in China as they do in the US. Whereas something like a social network has a lot of other dynamics. One being language, others being user behavior and things like that. I think it’s still early days but the big golden goose that a lot of people are focusing on is International Strategy.

Will:

How are you looking? I know you’re looking at investing in Chinese companies. What sort of main considerations you’re making there when you’re considering these limitations?

Adam:

I think what we’ve seen is that the use cases are hyper localized today. We do early-stage investments. International expansion is very far from the minds of most of the founders. Investors like to dream about that, but I think in reality it’s something several years away. It’s really about executing on a local scale, with the ability to leverage that technology asset and data into other markets when the time is right.

JS:

It depends on the market. In a market like China or Japan or Korea, you can’t operate as an American company or Canadian company, coming and hoping to build a business here, unless you’re working with local people. But you could be a Canadian company and develop a market base in the US or in some of the European countries.

Will:

Suji, we were just mentioning Japan and you have a unique perspective on that, being from China and having set up a company in Japan. Tell us about that.

Suji:

I will talk with the AI in Japan in general. Adam reminded me about Apple. ApplePay started to cooperate the Japan Railway and its payment system C card. I think Japan is on the cutting edge of entering a new generation and some revolution is happening there. There is also a large market. The country has one tenth of population as China, but the market is really huge. There’s some special and unique industry only exists in Japan, such as the anime industry. In the US they have game industry, but they don’t have an anime industry. I think there are a lot of chances. Technology is important, but it is not the only part. You need to find local partners, local business, build the local ecosystem to cooperate with them.

Will:

I really enjoyed John Sebastian’s talk. I thought it’s interesting looking at the case of Canada, because the government put a lot of money into fundamental research awhile back. That has paid incredible dividends for the whole AI industry. I wonder if you have the thoughts on how Canada can hold on to its AI leadership, when other nations are going to be looking and to follow. Everybody’s going to be trying to invest more and more. It’s going to be kind of a challenge.

JS:

It’s all about speed. Take the example of Element AI, we started the company in the fall and we will be in Asia with a whole stack here this year. We feel that’s the only way. We raised more money probably than what we should have based on the stage of the company. but the AI game is getting played now, and you need to move really fast. We’re serving more Asian companies now than Canadian companies, because the Asian market is more mature than any other markets in the world in our view, in wanting to implement our solutions.

Will:

I mean to that kind of point, do you guys think that there’s going to be an increasing talent pulled over to Asia? US or North America’s had this real center of gravity, but it seems like people are being pulled to Chinese companies. For example, there’s a lot of people come to MIT from China to recruit. I wonder if you guys would get a sense that that’s going to be a trend.

Wesley:

It was interesting yesterday’s talk from Jurgen at the beginning of the day. He talked about the idea about PhDs and true entries of PhDs and who you study with. In this field, you don’t have to spend quite a bit of time to really master it. Sometimes I worry that it is easy for everyone to come to the US or Canada, to study under certain professors and to get their PhD. They spend quite a bit of time and in their prime. They’re getting their opinions formulated a lot based in the US. It’s too much time. And once you are maturing in the US, it’s tough to go back later. That hope tree continues to propagate within certain countries. How you break through that, to me, seems to be a problem.

Will:

Adam, do you have any thoughts about that?

JS:

Yeah, I think Trump is helping everyone with that.

Will:

We will talk trump later

JS:

I don’t think people feel welcomed in the US anymore, especially smart people. I think we will see a dramatic drop in the US being the center of the universe of innovation. That will drive more of the people that have been successful in it to either go back to their home countries or to move to different ones. I think that will be a big accelerant. We’re seeing it already in Canada. It’s easier for us, because we’re just across the border. But as the immigration laws get even more restricted than what they are now, I think you’ll see more of it. In the end, why the US attracted so many people? First, it was cool, it carries amazing opportunities to work. You have Facebook, Google, Amazon, you have huge data sets. Obviously as a researcher or as an engineer, you wanted to go and work there. But as you now have amazing AI companies all over the world, there are great work opportunities here or in Japan, too.

Will:

I’ve been talking to researchers who have come back to China. A lot of them cite the fact that there is a lot of data here, and less restrictions on what you can do with the data. I think if you are a huge AI nerd, it compares a big carrot. Adam, what do you think about this?

Adam:

I don’t view this as US versus China. I think it’s the founders and the environments are going to grow up in the place. Market has a lot of other factors like the funding environment, accessibility to data. I think that different countries are going to either encourage or discourage that in different ways. I’m not talking about academic research. It is obviously important, but it’s also very much shared through the process of publication and peer review. I think the most important technologists who know how to use these tools are going to come from technology companies - Google or any other technology company, which hires most of the AI experts. Google will still be around. But there will be a small diaspora of these technologies that go out and start doing other interesting things. The same thing is definitely going to be true of Chinese companies. I think that those are going to be more of the dominating factors rather than rather than anything else.

Will:

Suji, do you have any thoughts you want to say about data?

Suji:

Most Japanese are more concerned about the user data. But it’s still a large business opportunity that they’re willing to pay more to not share the data with you. The Japanese want to research in AI stub as well. A lot of AI startups in Japan is found by Universities. One of the major investors is called UTEC - University of Tokyo Edge Capital. They found a series of AI startups and they share the data base from the University or even the government. Of course, you cannot publish it. But you can research with them together and they were willing to cooperate with you. One more thing I want to mention is the immigration policy. Japan just have released a new policy for the green card. You can spend as short as one year to get it. That’s better than what Trump offers, I think.

Will:

A lot of things are better than Trump.

JS:

I think a global problem but also regional problem for AI is that you have a few companies - here in in China you have Baidu, Tencent, and Alibaba, for example - that are aggregating most of the talent inside their platforms, and that creates a huge gap for the AI needs, for all of the other companies in the ecosystem. It’s true in North America as well, also in Europe. I’m not sure what the solution is yet but it’s something that we need to work on, because people will get stuck with only a few companies which have built really strong AI. All the other companies are forced to work with.

Will:

That’s a very good point. I don’t make these jokes about Trump. But one of the phenomenon that led to his election was the question of jobs. Jobs are disappearing in the US, and there’s a growing meme about AI and automation taking away jobs. I wonder if you guys have any advice for companies outside the US where you want to do AI work. How to address that? Is it a question of how you package and present this? Or what’s going to be the strategies? It is going to become an increasing issue, I think. Anybody wants to take that? That’s kind of a tricky one.

Adam:

I just want to clarify that question. Do you mean for people who want to start technology companies, or for people who are trying to deal about their jobs?

Will:

I ask for people who want to start a technology company in the US focused on AI. How do you deal with that? Because the jobs issue is a real hot-button one, right?

Adam:

Are you referring to what he just said about not being able to hire the talent? Or you were referring to just jobs being about public.

JS:

I think what Will is talking about - correct me if I’m wrong - is that AI companies don’t create jobs. They kill jobs.

Will:

That’s right.

JS:

That’s the mindset. I think it’s true to a certain extent. But before all the jobs are killed, the jobs will go to AI companies, because these companies which implement AI the fastest and the best will or more likely to be successful. As a result, they will hire more people, but they will destroy a lot of jobs at the same time in the markets, where their solutions get deployed. This will start in their home market. I think as a society or as society everywhere in the world, we need to think about what happens when half of the jobs in the world go away. because that will happen. It may not be in five years; it could be in ten years; it could be in 20 years. But when it does happen, we need to rethink the economic model. I mean, that’s something that if you’re an AI company, you need to think about. How you will be addressing that? Governments in general also need to be thinking about that, because that’s not something companies will solve. We need to reconsider how we define ourselves right now. A lot through what we do for a living will have to change in the future.

Wesley:

Really affect some of the transportation industry and the manufacturing industry, where you have quite a bit of jobs. Can AI really affect the jobs in those areas?

JS:

Within 15 years there will be no more truckers, which is the biggest creator of jobs in the US. We already have automated trucks and they are on the streets or on the roads already. What would you do with those truckers?

Will:

Although Auto and others will claim that the single truck driver can go further. It’s just how you spin it, or whether it’s actually an unanswered question? Adam, you look like you wanted to say something.

Adam:

I think it’s the job of the technologists to be able to make sure other people are informed and have the tools that they need to make these policy decisions. I don’t believe that it’s within the purview of technologists to be able to say, “this is the way that we should treat income” or “this is how we should tax robots.” I think that’s it’s squarely. I’m trying to pass off responsibilities to somebody else. I think it’s our job. It’s The technologists’ job who are developing this to make sure that the policymakers and economists are educated enough to be able to make those decisions.

JS:

I’m not sure whether I agree. Please let me explain. I don’t know as much about China. You look at Google and Facebook with the internet revolution for example. Trump in a part was elected because he was able to use Google and Facebook as propaganda machines, because they were not regulated in any way. The politicians did not understand the power of those platforms. In a way, even Facebook and Google themselves did not understand the power of their platforms, they weren’t forced to think about that. I think, ayahs we are going to have same impact if not more of the Internet. We need to take the responsibility of understanding what the social impact of our companies are.

Adam:

Just one last response. I don’t blame policymakers for that. I blame technologists for not sending it as a high enough agenda, to be able to communicate the types of things that are being done. There was a Peter Thiel quote “where if a company tries to convince you that they’re not a monopoly, then they’re probably a monopoly; and if they try to convince you that they’re a monopoly, they’re probably not a monopoly.” I think the same thing goes with this. For the technology companies, it has to be very high priority. I think it’s unrealistic frankly. Most of us in this room don’t fully understand the ramifications of what’s being developed. How would you expect a politician to? I think I don’t necessarily think that it’s right to point to those silly policy makers. I think that we need to 100% agree that we need more collaboration. It’s around educating them and giving them the tools that they need.

Will:

What’s the scene like in Japan

Suji:

One more thing to keep in mind is that people no necessarily want to work. I mean once the in Japan that Yamato used to hire a lot of people to take at mails and package by hand and now they use machine and business solution provided by AI stub booting so these guys who don’t have job now they go to actually do something more meaningful they have a better I mean to provide some the government provides them something to do like to go to university for adults and even have an insurance for that. It’s more of a problem for the society, not for the AI stubs. the China rate the average of preparation is still very young so we won’t encounter that problem fifteen or twenty years but they say we’ll come that people are getting elder in the other and they don’t really want to get to work and they want to find something more meaningful than repeat their daily life they work

Will:

yeah I think sorry Adam go ahead

Adam:

just thought of one other comment around job displacement I think like kind of one silver lining that from the US perspective of the manufacturing industry getting hollowed out is we have we have frankly very recent scars of an industry that got basically gutted and so one would hope that that that we have long term enough memory to be able to take lessons from that and say you know what did that caused and we can make speculations about kind of in the political environment but what types of people got angry but I think that you know we’re right to hear those voices now so that’s something that I hope that we can keep in our mind and learn from

Will:

I am I’m hopeful that there it is that the that trend potential trend may also create a lot of opportunities in terms of opportunities for retraining people and I think maybe AI can could play a role in that I think we sink some education-related AI startups I think they could do do good things so slightly more cheerful topic let’s do let’s just talk about China’s AI I seen what I this is slightly selfish because I’m writing a feature about it and I but I’d love to get you guys all of your thoughts on what you know house where it’s strong what it’s focused on just any thoughts on it so Wesley

Wesley:

one of the things I see you know if I think of consumer versus industrial one of the things that’s nice I’ve you know I’ve sort of been here three or four times over the last three or four years and definitely you know in this space so much of it is you it’s based in developing technologies and products for the home country and it’s not trying to think about exporting technologies so in this case I think while the I see the trend towards a lot of more consumer oriented machine learning or AI applications I’m really looking forward to seeing maybe an acceleration towards some of the more industrial ones in the whole IOT space I think it’s one that lends itself to you know data being very physical and local to a geography and I think it’s gonna help accelerate I think there’s great opportunity in China to kind of leapfrog and take advantage of that as opposed just focusing on consumer markets

JS:

yeah I mean I think another area where China has is very well positioned is it there’s less regulation for local companies in in in how to use consumer data so that way you can get a lot more innovation when you don’t have to worry especially for it in healthcare I think the other areas is it’s very difficult for a foreign company to come in and do business here so that will give local companies you know a couple of years to sort of figure things out before some of the large US or Canadian or European companies can come in and play with the local data center

Wesley:

just also really quick I mean just in terms of data centers I mean I don’t know how many companies have tried to build you know data center or cloud with it from an outside perspective you just can’t do

JS:

it impossible

Wesley:

so in this case with big data doing it out analytics you know it’s gonna tend to make it a great advantage

Adam:

I think we’re from kind of like a foundational level there are a lot of companies that have been built in China that I would argue probably just couldn’t really have been built anywhere else specifically in the US so I think you know always like I kind of look at it as a you know the banking system in the US for the past like a few decades has been good enough whereas you know the kind of the lacking in terms of from a customer standpoint like the locking in the banking industry was really one of the things that prompted such a high development of like mobile wallets in Asia which is you know it’s leaps and bounds ahead of I don’t know how many times you guys have tried to use like Apple or Android pay in anywhere but it’s like hard and I think that those types of factors are not to be not to be understated just because those types of applications that are built on this newer technology a lot more information is digitized the e-commerce revolution in China being much larger than it is from a user perspective how much time people spend online shopping their basket size their percentage of income spent on e-commerce those are all indicators of an economy that’s basically just grown up in a different environment from a technology perspective

Suji:

I think Chinese customers it’s actually more open to new products and companies they’re willing to try new business solutions they’re willing to give their permission to do something new and seems really hard in Japan but the other thing different is that China is actually they have a booming for like entertainment so I’m doing more in entertainment so AI plus entertainment so Chinese customers playing more and more in games and different stats that you have AI its class(?) in that area such as Meitu they have Beauty pass such things like that and this scene exting is very unique and only exist in China should be exist in China for a long time and Japan is probably similar but more difficult yeah

Will:

yeah I think it’s a good point and Adam I loved what you said about the WeChat I think I’d love to have something like that in the US but the fact that we have existing banking insurance systems payment systems means that it’s just my work which is real shame

so yeah so I found in my reporting that the scene here that the excitement the public excitement for AI is enormous and also the amount of money flowing into startups is crazy and it’s similar to the US but maybe even more and I wonder if you feel that there’s a bubble growing and if not AI winter is there gonna be some sort of fallout from that I must be a concern

JS:

yeah I mean to me the if you look at the internet you know that there was a bubble in 99 2001 but it was just ahead of its time right I mean the there was a reason for the excitement I think the difference with AI is AI gives you a like everyone’s connected already we’re all on internet that the datasets are there there’s a clear ROI today with the existing technology so I really don’t see it as a bubble I mean I think I think that the entire world is gonna morph and go on AI right every application will be AI driven every product that we use every service that we use like our society is gonna be run on it so it will happen and I think it’s we’re still very early

Wesley

the only thing I would say it just doesn’t maybe I don’t think it’s gonna be in a financial bubble but more of a bubble of the terms you know III think back in terms of like even the IOT term right that was probably a real big you know boom about 15 10 15 years ago with the RFID etc right and it didn’t really affect you know the financial markets as much but it’s just from a term it kind of bubbled up peaked and then kind of went down I think you know we’re in that little bit of a hype cycle around the terms AI but I don’t think it’s gonna have a big financial crisis association

Adam:

I think just based on the fact that the four biggest companies in the world have brand themselves as AI first companies it’s definitely here to say but with that said when people who people get that excited about a technology a bunch of companies get started that shouldn’t have gotten started and inefficient capital gets allocated in in inefficient ways and it goes away and the market corrects itself but from a technology standpoint this stuff is pretty I can say pretty confidently if I’m wrong you can point to this clip but pretty confident that it’s here to say

Suji:

I think it’s overheated but it’s not a bubble most AI labs including large corporations labs they’re trying to find something can revenue can generate revenue in the future but they are not of sure that why is the future come in at five years or ten years but I think they’re I mean losing in competition they have it there’s a too fierce competition in China in the u.s. right now I think five four stops doing the same areas and this affect each other so it’s better to find a new market use the technology you have and try to find something else and try to get into the ecosystem you can get better so it’s not a bubble just need to find the right revenue model so in Japan the competition is less fierce I don’t know about other cultures probably Canada or Russia is better place and yeah if you’ve got technology and try to find the local pattern for you find the generator find the right business model then you can it’s not bubble

Will:

ok that’s great well um we’ve run out of time and I’m sorry to and don’t a slightly negative no but thank you guys so much for talking that was really interesting and I’m sure everybody found that fantastic so please say thank you to the panelists here thank you