Welcome back to our course on FinTech Foundations and Overview In this session, talking about FinTech technologies, we're going to look at AI which is artificial intelligence, DSS, decision support systems and automation, various types of automation including artificial intelligence enabled automation And as we look at this, these matters for FinTech we have with us Professor Hilton Chan, Adjunct Professor at HKUST and CEO of his own company in the FinTech space As you think about artificial intelligence decision support systems, and automation as technologies how do these technologies influence the developments of FinTech markets? >> Yeah, in fact, a lot of these technology tools have been used in the past in different industry It's just that finance and banking is a regulated industry And in the past, data is not available When data is not available, even though I have the best AI engine, without the data right, the engine won't function It's like a car without gas, right? >> Right. >> And now with so many open-source data and also with a lot of data cleaning tools, a lot of data and tools I'll be able to sort out the data in a way that I could fit it into the AI engine So that an AI engine could do the processing to come up with something that has value, could help me to do decision-makings Or they could make the decision themselves >> Does increased computing power play a role in AI and DSS and automation? >> Yeah, definitely, definitely The hardware improvement and also the memory capacity increased in past years actually provided the infrastructure for AI and DSS >> Now I've been around awhile When I started learning about computer science as a young man we used punch cards >> [LAUGH] >> And we had mag tapes, and this was kind of a long time ago >> Yeah. >> So I'm revealing my age a bit But I was also, I was young when I started college, I started at 16, so >> Yeah, we did that in primary school, right? >> Yeah, I was using punch cards in high school when I was 12 and 13 So I don't know, we did some advanced things And both my parents worked for IBM at the time, and so you know grew up with technology But we were talking about AI back in the 1970s >> Yes >> We were excited about AI and we thought it would change the world But nothing happened It was much to do about nothing >> Yeah, I think a lot of the machinery in AI, firstly is a lot of those models at those days, it depends on computing power >> Right >> If the computing power is not ready, it will take you weeks >> Yeah >> [LAUGH] Months, to come up with results, especially in the finance market >> We used to say- >> What would [CROSSTALK] is that if you tell me the market is going to increase two days later? >> Yeah, we used to say we could predict the weather tomorrow >> Yeah. >> The only problem is it would take two days to run the analysis >> Right >> So you would know the weather yesterday, as predicted from two days ago >> Yeah, exactly, exactly >> Yeah, thanks >> Yeah. >> But what's happened is computing has increased in power by doubling every 18 months or so So what used to take 2 days now takes one day, then half a day, then 15 minutes >> Right, and also it's getting your smart phone, which is your personal device You don't need to >> You've got a super computer in your hand >> In your hand, right? Some people even saying that is part of you, electronic organ, right? [LAUGH] So it is part of you that is next to you that helps you to make decisions right, and do things, right? >> So when you get a new phone, that's like a transplant? [LAUGH] >> [LAUGH] So basically the interesting part And also, the other part is, even though you have computing power, but let's say you don't got field Data is the field So when we have open-source data with tons of data, right? >> So we didn't have enough data then >> Yeah, in the old days, think about it There's no Internet, no web, and then the only data we could get is through EDI as well, electronic data interchange There's no APIs >> Yeah, yeah >> So things are not in real-time So things are very slow, and then it's a download it's a backup thing that- >> So what's happened? We've got more computing power What's happened? What's different? What's new? What's amazing about AI today? >> Now, I think the technology, the computing model a lot of the mathematical algorithm already exists in there Now actually, we could implement it And then we have the data available so it would help us to make better decisions than before For example, let's say if I built a robo trader, an algorithm Now, in the old days even if I could come up with an algorithm I may not be able to get the data from the exchange instantaneously through the APIs And I don't have the computing power, I don't have the storage space And then a lot of these, then even though after let's say, obtaining those data doing data conversion, and then putting it into the machine, crunch it, and then get the result out, then it's gone The thing's already happened, right? But now it's not I could really get it up to speed I could do the prediction kind of before the market actually happen, right? So- >> Let me mention something I find fascinating about AI because I've worked with expert systems of variation of decision support many years ago I've worked with AI and people said, you know we'll be able to get chess right because it's a brute force game And we can do all the analytics But Go, the game with black and white pebbles that's popular in Japan and Korea, sometimes in China, we'd never be able to do that with a computer Because it's too complicated with too many variables And yet recently, AlphaGo beat the world's best chess players >> Yeah. >> I mean, the Go players, in the world And that was pretty amazing And then, AlphaGo 2, the next generation of that, without any programming self-taught, self-learning, within less than a year and a half, was able to beat AlphaGo, hands down >> Right >> In almost no time at all It learned itself That's new >> Yeah, okay, think of, okay I can even propose something called a very stupid algorithm But still be able to beat a human, for example If I've enough computing power, I have enough storage, and my IO time is fast enough, basically, I store all the possible moves >> Right. >> I store just store all the possible moves Whenever you make a move in my database, I would have to sequential three or four steps after that, then I just saw everything >> But the interesting thing about AlphaGo is it doesn't do it that way [CROSSTALK] >> They are much, much smarter >> It's much more sophisticated and much more adaptive in its ability [CROSSTALK] >> In fact a Korean master who was beaten by AlphaGo said that it cheats >> [LAUGH] >> It doesn't play the way normal people do It doesn't play like a human being It changes the rules of the game And you can teach AI now almost any game, including stock trading >> Yes. >> We got a mathematical genius that developed algo trading models, got himself incredibly rich >> Right, right >> Doing that, using better decision support systems for better trading And that was proof of mathematics and programmable trading beating the best hedge fund managers in the world >> Yeah, people use AI and algorithm in horse betting >> Yeah. >> In sports betting >> Right >> In a lot of areas Even though people used to be doing weather forecasting they just turn it into why not stock market forecasting, right? >> Why not? >> Yeah, so. >> If you've got the data might as well do the analytics >> Can't tell you >> What about automation? I mean, we've been automating things for a long time, since 1950s in finance We automated check cashing We automated credit cards, ATMs Described by Paul Hooker as one of the biggest FinTech applications in the history of banking, it's automation It's automation of bank tellers What's being automated today using these AI tools or DSS, which is new in finance?
