Jannine Krish:
Okay, let’s get started. Thank you everyone for joining us today at the Flybits digital banking series. Today’s event is called how top banks are building data ecologies that transcend the new normal and challenge big tech. It should run about 45 minutes to an hour depending on how the conversation and questions flow. And if you have any questions, please pop them into the chat box and we will save them for the end where our guests will take all of the questions. So without further ado, I’m going to pass it over to Flybits founder and CEO Hossein.

Hossein Rahnama:
Hi everyone. Thank you very much for joining us today. My name is Hossein Rahnama, founder and CEO of Flybits. It’s great to have two people that I have a lot of respect for to join me today as part of our webinar and sharing their view about what the future will hold when it comes to leveraging data and reinvention of our society and what role banks can play in the new economy. I’m joined today by Professor Sandy Pentland from MIT and Tim Karpoff from Barclays bank. I would first ask them to introduce themselves, tell you a little bit about what they do and then we will start with the webinar. So with no further ado, I’ll pass it on to Professor Pentland to introduce himself. Sandy.

Alex Pentland:
Great. So I’m Alex Pentland. People call me Sandy, a professor at MIT helped create the media lab, which you might have heard of. More recently, the Institute For Data Systems and Society where what I do is I have a consortium of financial institutions and countries which are looking at problems like digital currency, data guided investment, new ways of handling currency that are more digital, safer and so forth and building sort of a new ecology as it says in the title here.

Hossein Rahnama:
Perfect. Thank you, Sandy. Tim.

Tim Karpoff:
Hi everybody. I’m Tim Karpoff on Barclays head of strategy, which covers a multitude of scenes from traditional corporate strategy, business line strategy as well as things like corporate development partnerships and the like. It’s great to be with you all. I’ve rarely had the opportunity to share my bedroom with 200 strangers right at the beginning, but here we are. COVID gives us all kinds of new opportunities. And I had the great pleasure of working with Hossein on some opportunities between our two companies and we’ll see how we do. So great to be with you.

Hossein Rahnama:
Thank you. Yeah. Two things that our colleagues didn’t mention. First of all, Sandy is recognized as one of the top 10 data scientists in the world. I think he was too humble and not mentioning that. And Tim also has a very interesting background because before Barclays, he worked with the Obama administration and the regulators in the US. So they are bringing a lot of wealth of experience and insights, and we’ll try our best to leverage those expertise and knowhow as part of our webinar. What I like to start with, and this is a very fluid conversation today, and we will start first by understanding and getting our attendees view on the current market dynamics and what may happen in our economy after the COVID-19 crisis is over. Hopefully, we are all waiting for that.

Hossein Rahnama:
Sandy just published a book, is in the process of publishing a book and coauthored with a number of his great colleagues called the Building The New Economy and you can look at that as part of the work in progress WIP at MIT. It’s a great book, very inspirational. And also he recently did a Ted talk and he mentioned that with each crisis, whether it’s a pandemic or war, there will be a need to reinvent our society. So what I would like to ask him is that if he can elaborate on that and where does he see the role of data and digital in that reinvention process. Sandy.

Alex Pentland:
Yeah. So there’s at least three trends that are coming together. One is that we’re seeing this centralization of data in the tech companies and a real reaction against that with GDPR and things with CPP and just the general sense that this is not the right way to do things. So people are looking for ways to have access to data that isn’t the tech companies, and that’s going to involve getting people to contribute their data in a more sort of cooperative way. Think about instead of depositing money, they deposit data with you and then you can use that to guide investment. Along the lines of investment, what I’m seeing is lots of the leading investors are realizing that this new regime that we’re in is different. It’s like quantitative easing on steroids.

Alex Pentland:
So you’re going to have actually high inflation. And my colleagues that measure these things are convinced that we currently have something like five, six, 7% inflation. The CPI won’t change because that’s something that dictates government payments. They’re not going to let that go off to zero. And interest rates will go up, official fed interest rates, but actual inflation will go up. And what that means is your investment strategies have to be aimed in different ways, more towards real assets will appreciate along with that inflation rate, that real inflation rate. And so what people are beginning to wake up to, and I see sort of a herd going on this way is more investment in community infrastructure, building buildings, building businesses locally. This is like the McDonald’s thing, right? You have this thing, but actually the way you make money out of it is the land that it sits on.

Alex Pentland:
Walmart has the same strategy. Yeah they sell things, but it’s the land investment that they’re after. And that type of strategy is going to become something that is the winner. And you’ve seen this in different parts. I’ll talk about what the Swiss have done for instance, what the Singaporeans are doing and so forth. But that’s really the game changer and banks can play a role by being the holders of this data for investment, for data guided investment. It’s not the way the tech companies do it, where they want to own your data. I think that banks can be the place where you deposit data and then use it for the community and the community can see how you’re using the data, which makes it all okay under GDPR and other things and gives you insights that the tech companies won’t be able to get because people believing that they own their data and control their fate will give you a lot more data than they’re going to let the tech companies hold on to. So that’s a little bit much all at once, but there you are.

Hossein Rahnama:
That’s great insight. Thanks Sandy. Tim, you have an executive role in a global bank on a group level that this institution has seen a lot. It has been in wars. It has seen the industrial revolution. It has seen financial crises. How do you see COVID-19 or the post COVID-19 era playing a role and how the bank is adapting to that and transforming itself? The reason I ask is that before COVID-19, as part of Flybits, when we were working with banks, almost all of them had something called digital transformation. It was a strategy and they were supposed to deliver on this strategy in a matter of years. And now we are hearing that now they have to do that in a matter of months. So do you see that dynamic playing a role in Barclays and how does your organization look at that?

Tim Karpoff:
Yeah, I think it’s an excellent question. And I think the point you raised regarding the accelerated needs for transformation is exactly right. But to some extent, I think that that is a tactical factor that is created by the pandemic. And I think that the phenomenon that Sandy identifies is actually much more important longer term to the health of the institution and where big banks will make money, because it’s much more fundamental to the role of banks in society and it’s going to create real headwinds, I think and I’m paid to be a pessimist. So I will continue in that vein at least here.

Tim Karpoff:
But if you sit back and go to first principles that Sandy identifies, if you’re in the business of intermediating between people who have got capital that they’re willing to deploy and people who need capital, and you have an interest rate environment that quite frankly, it looks really screwy and the central banks that are deploying a lot of money, it’s really hard to make net interest margin. And I think fundamentally that’s going to be challenging for a while, which makes banking, the process of money creation, asset liability management, et cetera, a tough business.

Tim Karpoff:
European banks have faced it for the last couple of years. I think US banks are going to continue to see pressure, probably not to the same extent, but if you start with the idea that banking is predicated on the idea of a positively sloping yield curve and effectively you have sort of the first, you have the function the first and the second derivative, all in a rough place. And by that I mean, rates are low, the yield curve is not positively sloping and there’s no expectation of change in that yield curve, it makes banking tough as a business. So I sit there and I think all right, banks are sitting on all this wealth of data. They’re going to have to find out a way of how we create value for customers and clients that I think can in many ways be enabled by the ability to maintain and enable users to put their data to work as a trusted repository. Probably not in the same way that technology companies do, but as that repository.

Tim Karpoff:
How will they get monetized? I don’t know today. But I think it’s going to be necessary for banks to figure out other sources of revenue, because I think NIM is going to continue to be under pressure. To your initial question regarding what digital transformation looks like, I think COVID didn’t for the most part create new trends that we hadn’t expected. We just expected them to play out over three, five, seven, 10 years. Now they’re going to play out over three, five, seven or 10 months. And the ability to move more quickly, be facile with technology and data, and use that as a way to accelerate transformation is just going to be critical. But I actually think the longer term issue, the one that Sandy identifies is the bigger one.

Hossein Rahnama:
Interesting. One of the things that a few months ago, Sandy and I were discussing was that up until now, we have done a lot of good work and especially big tech had an important role in helping us to build a very strong communication medium globally. The fact that we are sitting here and have a multi point high definition video call was not something that people could imagine 10 or 15 years ago, and now we take it for granted. So when we were thinking about the future, we said, “Well, on top of this communication medium, I think there is a need for a new transaction medium.” And there are different elements of that are available now from blockchain that allows us to decentralize data from security protocols. Some of the great work that Sandy’s group is doing around algorithmic transparency and the protocols such as opal.

Hossein Rahnama:
So the question that I have for Sandy is that, how do you see this transaction medium is getting formed and going back to what Tim mentioned, do you think this could be a vehicle enabling banks to really monetize data? And instead of primarily focus on money and financial assets as the unit economy, they can start leveraging data as a principle of network effects and generating value.

Alex Pentland:
So again, it’s somewhat complicated, but there’s a couple of main themes. So one is this notion of data as an asset. So the return on investment in terms of dollars is going to be moderately pathetic. On the other hand, services like helping manage data for investment could be quite lucrative and the appreciation of real assets, which the bank might not want to do themselves, but would be willing to advise and help and mediate, that should be a pretty good business. There’s a lot of the world that’s going to need to be rebuilt. A special thing that most banks have is they have community centers. They’re called bank branches. Currently, they’re not very interesting. They’re more of a loss center than a profit center, but you can imagine that being the place where communities plan investment and interact with each other and people manage their data resources as a community there.

Alex Pentland:
I think that would be a very, very profitable, useful type of thing to gave you data resources for guiding the investment that nobody else would have. The third trend, which I think people are insufficiently aware of is digital currencies. So Bitcoin craziness, right? Libra ha ha ha ha. Technically, Libra wasn’t bad, but the governance was like crazy. However, in the last little while, last couple of weeks even, right, you’ve seen the monetary authority of Singapore with the investment of Temasek declare that they are launching their digital currency. So Temasek for those of you that are aware owns 20 to 40% of almost everything around the Indian ocean, South Asia, Africa, places like that and China has done the same. They’d announced the Belt and Road is going to be based in their digital currency.

Alex Pentland:
The thing about digital currencies is that they’re dramatically more efficient. You don’t have to redo KYC everywhere. AML becomes automatic. It’s just some computer stuff running in the background. Transaction costs change dramatically. A lot of the traditional profit centers like a bank are got to get nailed. Okay. On the other hand, the reach of these systems will make new sources of profit from volume, from new projects huge. And banks have to realize that all this BS about digital currency is becoming real because now you have serious banking entities having declared that they’re going to transfer their trillion dollars worth of investment into this format. So it’s beginning to happen. And that will be something that we need to in the West follow along and use for local investments, data guided investments, smart investments. And that requires working with the community to be able to get those assets and do the sort of planning and asset guidance that will generate the profits for the future.

Hossein Rahnama:
Interesting. And on that note, the question I have for both of you and we start with Tim is that where do you see the role of big tech in that? I mean, we have these massive centralized data repositories, and if you consider data as an asset, you can consider them as asset repositories. What role they can play in the new economy. When it comes to a bank, will a bank play and cooperate with them and share data in a privacy preserving manner, or the bank will position itself as a competitor because data will become the new asset. So perhaps I asked that question from Tim because it’s every day in the news that one of these big tech companies is getting further and further into territories that was more or less the territory of a bank. And I would love to hear from Tim considering that he has an executive role in one of the most important banks in the world. How does he see that? And then we’d love to hear from Sandy on what he thinks the role of the big tech could be there. Tim.

Tim Karpoff:
Sure. So cutting to the chase, my expectation is that all of the big tech firms will get further and further into financial services to the degree they can walk up to the line of being regulated in a prudential fashion, whether it happens to be as a depository or as a broker dealer, because it’s not fun and I think they’ll want to avoid it. However, they’re not stupid. So they will do everything possible to redefine that line as far as they can further and further over so they can provide as much of the service without being regulated. And I think you’re already starting to see that in one of what I would say are the three functions of finance.

Tim Karpoff:
And if you boil it down, finance is basically three things, right? It’s payments, intermediation, it’s risk transfer and it’s matching people who want to borrow money and who want lend money. And whether that can be on a bank balance sheet, capital markets, whether it’s raids, FX, whether it’s using allowing people to withdraw from an ATM, deposit a check or you Swift, right? It’s all sort of a form of payments intermediation. Technology companies have moved very far to push the envelope just as Sandy just noted on payments intermediation, whether it happens to be new networks, whether it happens to be digital currencies, the manner in which one moves an asset from one account to another account has changed dramatically over the last 10 years.

Tim Karpoff:
And in large measure, not because the banks push it to move dramatically over the last 10 years, but because somebody else pushed it to move dramatically over the last 10 years. And I think you’re even seeing in the United States no further and further push from the tech companies and from fintechs for relaxation of certain rules to allow them to play greater and greater roles. You see in the OCC in the US, there’s a look to sort of expand what is a charter for these institutions so you don’t take the whole of banking regulation, but you can participate particularly in payment schemes.

Tim Karpoff:
In terms of the relationship between us and them, I think it’s evolving. Obviously, a number of them are some of our biggest vendors. And I think for all of the major institutions here, they buy services from these companies and also they treat them as clients. We want to be able to provide loans on Amazon, right. We want to be able to facilitate the provision of transaction accounts, maybe next to Uber, right. Whatever it happens to be. So they’re both client and they’re a vendor to you. I think the trick for all of us as a financial services institution faced with big tech players is we can’t outsource it all.

Tim Karpoff:
Some part of the data and technology competency, it goes with being a financial institution is going to have to be our sweet spot, something where we differentiate ourselves. And we’re going to have to be competent at building our own software, thinking about how we maintain data, how we use it. And we can’t push it all out to the tech companies because they’ll eat our lunch if we do that and we will eventually become just dumbed down and chewed the end of the day. Where that line is exactly going to exist, I don’t know. I spend a lot of my time thinking about it, but it’s figuring out what that line is that it thinks going to be critically important.

Hossein Rahnama:
When it comes to the synergy with the big tech, what we have observed is that the way a lot of banks used to think about data-driven engagement, their view was that, okay, we create a centralized repository of data, many call them a data lake, and then we can do all sorts of things with that data and then we push information to our users. And then we started to see what happens when you do that, when it comes to privacy issues and fatigue on digital channels. And I would love to ask Sandy what the role of regulation could be there, because let’s say when you look at open banking or the premises around that, or PSD2, basically their model is like the data is owned by the user, but the experience provided by the large entity should be so good and so useful that convinces the consumer to share data.

Hossein Rahnama:
And with that naturally, there’s going to be a new physics of data in which in the past it was, hey, I know a lot about you and without any transparency I can use that to generate interventions, into a model in which, hey, your service should be so good in order to convince me to share data with you. Sandy, where do you see that? Is that something that banks can learn from big tech and combine it with regulation and decentralization to create a new service model for their customers?

Alex Pentland:
I think this notion of decentralization is really key because I work with a lot of the very senior regulatory bodies. This notion of big central data is dead. That’s not going to be going forward. And also, I mean, I’ve been on the advisory boards for Google, for AT&T, for Nissan, for Motorola and I can tell you something is that if it isn’t a sort of Google like process, Google is terrible. They are not invincible except in their area. One of the things they don’t know how to do is localize things. So this notion of local communities having like sort of a data union or just like the credit unions of the past, or the way for instance people have localized retirement plans and things like that, that’s very un-Google. That’s very un-Microsoft. And they as a cultural thing would have a hard time plus the regulators won’t allow them to do their sort of normal tricks there.

Alex Pentland:
I spent an hour with the CEO of Barclays about a year ago at Davos talking about branches and about the notion of making all of those branches into a community investment center. So not just the money of the people, but can you actually use those community centers for outside investment for the big funds? I guess we have a lot of the Goldman Sachs on the line here. So can those community centers help Goldman Sachs invest in things in the community with the data of the community and with the ascent, the agreement and knowledge of the community. And I think that’s quite possible. And it’s a unique cultural attitude that banks have is that notion of local service and local communication that the big tech companies don’t have. And then all the regulators are bound and determined to make sure that that’s the model going forward.

Hossein Rahnama:
That’s interesting. So the question I have for Tim, and this is something that I have been collaborating with Sandy and his team for quite a while, which is okay, if you go to a community almost you can find bank branches and a lot of banks over the past years thought about oh, the branch of the future, how does it look like and put digital signage and everything. But when you go there, you’re like, “Okay, what is special about this place?” Especially now that things are becoming more and more digital. So is it foreseeable to see that the branch of the future will actually become a hyper localized data hub for that community, allowing SMEs and community members to come together to share that data and those insights will be used by community members to invest in their local economy?

Hossein Rahnama:
A lot of economists are saying, especially in places like the United States, the reboot of the economy is very much dependent on small businesses and entrepreneurs. Can that be a role in which branches can play and really transform themselves from essentially a financial vault for their community to a privacy preserve data hub for that community, with the purpose of rebooting that economy. So I would love to hear your view on that, Tim.

Tim Karpoff:
Yes and no is the short answer. I think that the fundamental concept of banks playing a role in local communities to facilitate economic development through enabling small and medium sized enterprise and utilizing bank data to drive growth, absolutely 100%. And frankly, Barclays runs the largest network of business incubators in the country. We call them Eco labs distributed across all of the UK and I think they do precisely what you’re suggesting, what Sandy is suggesting. Not yet at what I would call the advanced 401 level, but certainly at the 101 and emerging to the 201 level and been incredibly successful for us. And I think that will be a trend. And I actually think COVID enables that to some extent, because you have this move of intellectual capital back to communities from urban centers, which I think you’re already seeing. Obviously COVID will recede, you’ll have some return to work, but everything that we see in our day to day data suggests that in many of these communities with professionals now spending more time at home and in those communities, they’re more invested in their communities.

Tim Karpoff:
They’re more interested in small businesses. They’re more interested in spending money there and we think that’s a trend that’ll continue. The no part of my conversation is around whether it will be in branches in particular. I think it will definitely exist in localities and in physical spaces and those incubators can play a role. The trick with branches is … and there’s too many of them to maintain the quality that you’re looking for if you’re going to actually provide that service. So we have 900 branches in the UK. I think it would be very hard to get 900 of them up to the level that you would want. It would deliver the kind of service you’re talking about. I think we can do a hundred or several hundred of them, but that’s just too many. And if you look in the United States, JP Morgan has 8,000 branches. I think it’s unlikely JP Morgan will have 8,000 centers that are able to provide that service. But the fundamental idea I think is the right one. It just may not be that particular channel through which it comes.

Hossein Rahnama:
Yeah. Yeah. That’s interesting. One area that Flybits is doing a lot of work on is to really enable and empower banks to form these data alliances. So this idea of bunch of corporate entities coming together and share data and going after the same set of customers, it’s nothing new. It has been there for many, many years. And what we have done recently over the past two years is that we allow a bank to become a hub in the middle and bring a telecom operator, a supermarket and energy company, and airline to all get together to share data and then bring value to their customers.

Hossein Rahnama:
And it’s interesting, the observation I have had is one of my recent examples is that a government and a university network in a country came to us and said, “Can we leverage the same protocol that the bank is using for transferring money, and we want to use it for enabling students to transfer academic records? So instead of they go and picking up their transcript from one university and deliver it to another one, they use let’s say an e-transfer protocol to exactly do the same thing, but it’s being used through a bank.” We are having indications that the same thing can happen with medical data or EMR data. And instead of really hosting the data in a cloud somewhere, very similar to how you host your photos or sometimes music locally on your phone or some people refer to it as the edge of the cloud.

Hossein Rahnama:
We can use that capability to leverage the bank to really become a conduit or a proxy to share this data. We are already seeing that through three projects that we have globally, and I would love to get your opinion on what do you think about, and what is your view on the formation of these new data alliances? Because we saw that in the airline industry. In the eighties when airlines had problems, they started to form alliances. And what do you think forming these alliances between a bank and the telco and a loyalty program and a travel entity could be a response to big tech in which the data can become more controlled, more privacy preserved and then these alliances will start to be a competitive force to centralize the big tech. So maybe we start with you Sandy and then we go to Tim.

Alex Pentland:
Well, among the people that support research are several such alliances that are interested in the telco, the government, typically as part of it also, the bank and sometimes the medical institutions. So for instance, we’re setting up our project to actually operationalize this in Australia. We’re just beginning something in Luxembourg. We’re doing prototypes in some of the developing countries, Senegal, Colombia and some of the US ones. The US is a little slower on this because of the amount of discussion that needs to happen and there’s been the same bubble of urgency. But when you get to this point of, you have to now reconstruct the economy somehow, things are in tough shape. How are you going to do it? The key thing is data, who has data? Well, banks, telcos, government, medical, they have to come up and put things together.

Alex Pentland:
Some of the interesting things we’ve been able to do a prediction of store sales for stores that don’t yet exist. And it’s as good as the predictions of sales you can do for stores that have a long history. Well, that has real transformative effects on credit ratings for SMEs. We’ve been able to do the same thing for sort of government investments. And that sort of data-driven technique is really changing the territory because all of a sudden you can directly relate data to return on investment, but data gives you X percent more. I mean, it’s not a small percent, it’s 10 to 20% more, well return on investment and that’s something which people just have to pay attention to.

Hossein Rahnama:
Thank you. Tim, we’d love to hear your insights as well on that.

Tim Karpoff:
Yeah, I would say I generally agree. And I think that there are natural synergies between banks and other institutions, particularly airlines is a good example with airline rewards programs, anything with a loyalty scheme, but something that is effectively a quasi financial asset, right? And to the extent that you’re able to come up with some type of tool that allows people to aggregate their different types of assets into some type of multi-asset ledger and get people an entire look, I think that could be quite powerful. And the ability to tell people you can maintain your data in the variety of different institutions which are the primary interface for that particular service, but we’ll give you an ability to aggregate it. And by the way, you don’t have to worry about somebody controlling all of that data in one place to go back to the centralization point Sandy identified earlier, I think that’s quite powerful.

Tim Karpoff:
The challenge of it is I think for that to work and we’re spending all the time, both on the consumer side but also on the wholesale side, the organization and the manner in which you keep your data needs to be consistent and relatively well standardized to facilitate that type of interaction. And that includes not only within a particular institution, but there needs to be a degree of common definition across various institutions to facilitate that kind of horizontal consistency. It doesn’t exist and I wouldn’t say it doesn’t exist just at Barclays. I would be surprised if at any of the major financial institutions there were working on it, but especially as you have institutions that have brought together multiple different institutions over many, many years, have multiple different products that have evolved on different technology stacks back to individual data libraries to which they are then trying to pull data into a common data lake on the backend, where you don’t have common definitions for the information that came out of the particular data stores, it’s a mess, right?

Tim Karpoff:
And so I think everybody is on that journey to try to get to a place where they just get hygiene first, which then facilitates the ability to do all these other interesting things. But that hygiene question I think is not quite done. And at least what I tried to talk to people individually is that’s great. We need to do these other things, we should keep going and we should have these ideas, but fundamentals enable everything else. Let’s get the fundamentals and then we can move forward to the other stuff.

Hossein Rahnama:
Interesting. Interesting.

Alex Pentland:
So can I jump in just for a minute? Because this is like one of my main sticks here is that having the same labels and the data lake is a profoundly bad way to think about it because it’s hardly expensive. There are no projects where that’s something that turns out to be cost effective and maintainable. But from a sort of computer science point of view, you don’t need data lakes. What you need is you need query mechanisms and you don’t need having the same labels. You need to have labels that can be translated and we’re beginning to see systems that do this. So for instance, I work with the European union, Eurostar, and they have this problem in spades.

Alex Pentland:
And what they’ve done is adopted this query level where data is transferred not at the individual level, and in fact, the data is not transferred. Queries are done on aggregates and you can’t do everything that way, but you can do 80% of it that way. And the requirements on consistency of label and quality of data drop dramatically, and there’s no need to redo existing systems. You just need to build APIs for the systems that are there. We call them as open algorithms because the lawyers get to look at what happens to their data and that means that the legal costs dropped dramatically too.

Tim Karpoff:
So I think it’s right in many cases. Where we’ve run into challenges, because we looked at much of the same thing is particularly around risk and capital calculation data. So to the extent that you are running model space calcs using a variety of different fields from that information, at least our experience has been prudential regulators are not yet there. They will get there, hopefully, but-

Alex Pentland:
The traditional models require this sort of consistency on individual level aggregation and what the research and the sort of pilot systems, the beta systems are finding is that you can do it in ways that are safe, but those aren’t the standard ways yet.

Tim Karpoff:
Correct. And given the capital calculations for bank and your return on equity drive, many of your financial decisions [crosstalk 00:38:12] and regulators have the final say as to whether they are comfortable with the method by which you are calculating those-

Alex Pentland:
Let’s work on the regulators.

Hossein Rahnama:
And it’s interesting you said that because we have seen that trend so many times on the internet, even from the music days, that first you’ll have mp3.com that everything was super centralized, then you got Napster that was interesting, but became a mess. And then we found the equilibrium, same thing happened to content. Same thing happened to even how we use social media these days. The last question I have and then we’ll go to the audience. There’s was actually a question online that I like to bring it up. I’m glad that we are really focusing our webinar on data before we talk about AI, but the question is about the role of AI. And my observation especially with banks is that we went to a phase that AI became this buzzword. The board went to the CEO and said, “We want some AI.”

Hossein Rahnama:
And CEO went to the CO, “AI, AI, what are we doing with AI?” And we observed many times that the bank went and procured their very large type of a AI system and really could not deliver a lot of impact for the bank. And the reason for that in our view was that even a subset of AI which is machine learning, there are about 55 different ways that you can use machine learning. So the context is going to be very important. And in many cases when we talk to banks, they are like, “Okay, in order for my AI strategy to scale, first and foremost, I need to have a data strategy. I need to figure out how I manage portability of data between my business units with outside the bank.” And that’s key because you have no idea how many times I meet with entrepreneurs, data entrepreneurs, AI entrepreneurs, and they’re all frustrated like oh my God, I had this amazing technology and the bank is not giving me the data.

Hossein Rahnama:
Of course, why would they? It’s a sensitive data. So first before we figure out a scalable AI strategy, we got to figure out how to move this data that is privacy preserve and then generate value out of that. So I would love to hear your view on that at Tim and Sandy on how do you see AI and how do you see the role of your data strategy when it comes to your AI strategy. And by AI strategy, I essentially mean that instead of prescribing your digital channels, what you do, you create an automation in which there is more context and intuition in your channel that can predict, that can associate people with each other or products with people or clustered groups of people together. What is your view on that Tim? I would love to hear your view because the bank of course has done some great work in AI, but we’d love to hear your view on that.

Tim Karpoff:
I would love to hear what that great work on AI is. If you could pass that onto me, it’d be helpful. I can go look for it. So look from my perspective, and I’m not a technologist by background, but I’m a mathematician by training, so I have at least a fleeting understanding. The challenge at least that we have internally I think is how do you actually put this into practice in a large complex, bureaucratic system that tends to be product aligned? To Sandy’s point, there’s obviously certain challenges by the manner in which we store data, but there are a lot of capabilities, strategies that can help you deal with those challenges even if you don’t want to spend a huge amount of money to go back and retrofit your data libraries, they give you a chance and we’ve done much of that, right? We have been able to cobble some stuff together that gives us a chance to actually look for insights, see what types of associations we can see.

Tim Karpoff:
And in a couple of places, particularly around our UK retail bank, we have been able to pilot a variety of different features, products into our digital app into the way that we put targeted offers in front of customers that have been interesting. I think the challenge for us more than anything else is the time horizon, right? If I sit back and I think about what are the benefits that potentially will come from being able to deploy AI at scale in a retail business, there’s the ones that I can sort of conceptualize as a strategist, whether it’s easier product origination, easier customer acquisition by lowering the search costs associated with either of those, better risk pricing. That all makes sense, but it’s the stuff that I can’t think about that is really where you’re going to make some money, but it requires patience, right? Because what you’re going to get out of it is probably going to be something totally unexpected.

Tim Karpoff:
And it’s going to take a fair bit of time before either A, you identify that trend that you didn’t otherwise see and then B, you figure out what to do with that trend and how you’re going to build your business around that trend. Right now, I’ve got my team doing a fair bit of work around trends in unsecured lending and sort of where growth is coming from. And you go back and you talk to the business and the first thing they’ll tell you is like, “Great. Let’s look at a product level where the people are taking it in revolving credit, they’re taking installment loans, point of sale. It’s like, I don’t give a shit.” Like, why don’t you just tell me the entire world of unsecured lending in a geography, who’s taking it, what’s the purpose they’re taking.

Tim Karpoff:
Cut it 18 different ways, right? And it’s just, the bank is not hardwired to think about this except in a particular way. AI can facilitate new ways of actually organizing information and new ways of actually organizing products or services that you can put in front of customers. But because you get pressure at the senior level really at a quarterly basis, at a yearly basis to make particular returns, as you’re thinking about investment, it’s really hard to have the time horizon necessary to allow things like that to come up. And for me, that’s the hardest challenge is how do you have the patience in the organization to allow these things to mature when they’ll clearly pay off, but I can’t tell you exactly when.

Hossein Rahnama:
Yeah, that’s great. Thank you. So Sandy to your view on AI, and especially when it comes to banks, and then we’ll go to questions from the audience. We have a number of them.

Alex Pentland:
Yeah. So I’ve been through a couple of these. I chaired AT&T’s big data board and we started with internal processes. That was one thing. So by year two, we were saved a billion dollars in operation costs just internally. But it wasn’t just the billion dollars, what you got is all sorts of people aware of the tools and just sort of how to think about it. They had something they were familiar with where they’d seen AI connect things. Another example is I sat with a board of Telefonica and they decided to do digital services as part of Telefonica. And it was a stone-cold disaster because of this culture issue that Tim just talked about. They didn’t understand, the culture just wasn’t there. And so in the end what they decided to do is do a spinoff, an allied wholly owned company which has some of the same DNA and has type relationships, but really is aimed at different sorts of things.

Alex Pentland:
So it would be like a data driven investment firm that is wholly owned by Barclays and you build that up as a separate but contributing business and then over time you build links. The second thing is don’t try to automate services. I think that’s a problem. AI is wonderful, it optimizes things, it conditions changes, it can break horribly. You saw Amazon next day delivery, my God, it’s a miracle. And then COVID hit and things broke massively, and they were able to salvage it, but only by cutting back all sorts of products and going back just to their real core, and they dropped the next day completely, you might’ve noticed and raised the price. So building tools for people to do systems that humans understand and are sort of in charge of is the way to build that sort of resilience to change, but also to build a culture around using these tools.

Alex Pentland:
I don’t mean to go on and on about it, but we’re all familiar for instance with weather prediction now. Weather prediction is amazing. You see 10 different AIs shooting off, shooting against each other on the screen. They show this corner probabilities where the hurricane is going to go, where’s the weather front. Everybody’s familiar with it, but what that is is not AI telling you what’s going to happen. It’s the AI models giving you competing interpretations that the human conveys right in that very human way. And you have to keep that human centricity because otherwise you’ve got this culture problem. It’s like, what’s this alien thing? People don’t trust it. They want to be able to tweak it. They want to be able to see competing things and that’s as it should be, because that actually results in huge risk reduction.

Hossein Rahnama:
That’s great. And this is why we always said even at the media lab that AI as a decision support system and not necessarily as a decision making system and really focusing on the symbiosis between the human and the machine to make better decisions. So that’s great. Really appreciate those insights. So I think we have a number of questions now from the audience and perhaps Jannine, I pass it to you to go through some of these questions. If we don’t have time to answer all of them, we will share the answers with you after the webinar. Jannine.

Jannine Krish:
Yes. Thank you Hossein. So I’m going to combine two of the questions because they’re, I think you can link up the answers and this could go to any of you. So would you say are fintechs enablers or challengers in this new economy? And the second part is how are banks working with fintechs to compliment their data investments?

Hossein Rahnama:
I think we can ask Tim to comment on that.

Tim Karpoff:
Okay. The first question, both enablers and challengers, and the second question I’ll come to a couple of examples. On the first question, there are firms I think like Flybits who really quite powerful capabilities that can enrich the service that from like Barclays or others that Flybits work with can provide to their customers. And in that respect, I think that there is a symbiosis by which it’s something that we want to add. We improve our relationship with customers. Flybits, I think gets greater and greater exposure to more and more people. It’s mutually beneficial for us to do business with one another. There are others where I think fintechs are more directly trying to compete with incumbent players. So in the UK, you think of the challenger banks whether it happens to be Monzo, Starling, et cetera, et cetera.

Tim Karpoff:
And I think the interesting issue there is twofold. I think one, people regularly ask me who’s the winner between fintechs and the incumbents and the answer is the customer because what you’ve seen is radically improved service for the end consumer, both from FinTech and from the incumbent, and quite frankly, spreads on certain products collapsing to zero, right? You take an example, Revolut or TransferWise and their FX services in the UK, people ask me like, “They’re making a huge amount of money?” I’m like, “No, it’s just that they push the price down and they’ve competed away the margin in the service.” And that’s probably capitalism, right? Like that’s what’s supposed to happen. Long term I think the question is really between the two sides an issue of whether the incumbents, people like us will evolve and start to sort out and get the questions that we talked about today both with respect to data, but frankly, with respect to our core architecture as well. And then frankly on the other side whether fintechs will be able to get scale particularly in the asset side of the balance sheet and that’s a good question.

Hossein Rahnama:
Thank you.

Jannine Krish:
Okay. So this one’s a bit of a funny one for Sandy. Did we waste their money and efforts building data lakes, Sandy?

Alex Pentland:
Yeah. It was a natural sort of a human process. We’re kind of like get it all together to sort of figure it out. But from a computer silence for a review, it’s actually not a very good idea and unnecessary. You need to have little data lakes to experiment. So it helps, but the moment you do that, you give a single point of failure, a place for attacks to happen. It becomes very hard to audit what’s happening to the data. And legally, it’s a problem because lawyers don’t want to see their data, their client’s data in your hands. They don’t mind in specific questions, but I think it’s just a natural thing. We used to have mainframes and you’d bring card decks and stuff. And then we had … and this is a natural evolution.

Alex Pentland:
But I think the pain that people have seen with data lakes, 80% of the money goes into data cleaning. You have to redo all of your existing systems. That’s crazy. And people are saying, “No, we’re just not going to do that.” And so this more distributed example which we call opal, open algorithms turns out to have huge advantages not just from the computer science, but the legal point of view, getting agreements among different parties, hugely easier.

Hossein Rahnama:
I think that’s a very important topic. I think Sandy done a lot of work there. Opal is definitely an algorithm that we looked at and then leveraging it at Flybits. Other big companies and cloud providers are going the same route. Some of you may have heard about Azure and their trusted execution environments. So if audience are interested, we would love to tell you more. I’m kind of wearing the selfish hat, but that’s also the hat that Flybits is taking when it comes to data orchestration, not to centralize things, because we have all seen the horror stories whether was Sony or Capital One and others. And these mostly happened because of human error, because someone just forgets to do a configuration. And then the whole, this is even the art of war. If you bring all your soldiers in one area and really would build a strong turf around, then it’s just a matter of time. So we love to communicate on that. It’s something that is very close to our interests. Jannine, we have time for one more question and then we’ll communicate with attendees if they have other questions.

Jannine Krish:
Okay, great. Thank you. So we have one more question here. This one’s for Tim, as we see open banking regulations solidify in Europe, it seems like more and more reality for us in North America. When do you imagine as the students we will see our own set of established regulations concerning open banking in America?

Tim Karpoff:
Well, as I think probably it’s pretty clear from my bio, my politics are well known. So I don’t think it’ll be immediately, but it’ll probably be in the second term of a Biden administration [crosstalk 00:55:10].

Jannine Krish:
Okay. Thank you. I think we’ll wrap it up at this point. We’re almost at 12:30. So I’d like to thank all of our wonderful guests. That was quite an engaging discussion. There was a few more questions that came in so we’ll also try and respond to those separately. And I look forward to seeing all of you attendees again at our next digital summit that will be in August. So stay tuned for an invite and I wish you all a wonderful week.

Hossein Rahnama:
Yeah. And thank you, Tim. Thank you, Sandy. This was great. I enjoyed the conversation a lot and I’m sure it was the same thing for our attendees. Thank you for your time and we’ll stay in touch. All the best. Bye.