We have talked about using portfolios to manage AI projects, we've talked about the democratization going on in machine learning. This is a theme that I've written extensively about with Apoorv Saxena, who was formerly with Google, and helped to implement many of the AI initiatives there, and is currently the global head of AI at JPMorgan Chase. These articles that we have written are available as optional readings as part of this course; I encourage you folks to take a look at it. Now let's jump into a conversation with Apoorv. Hello, joining me now is Apoorv Saxena, who is the global head of AI at JPMorgan Chase. Apoorv, thank you for joining us. If you can please introduce yourself and your background a little bit. Thank you Karthic for inviting me to this conversation. As you said, I lead AIML for JPMorgan. I've been here almost one year, and I fundamentally believe that AI is very transformative for finance. This has been my theme for the last few years. Before JPMorgan, I was with Google where I ran Google AI verticals team. Essentially, AI applied to healthcare, AI applied to finance, AI applied to industrials. We fundamentally believe AI is one common theme that runs around transforming multiple industry, and I'm making that happen at JPMorgan. Great. In fact, on that note, I would love to hear about how AI is having an impact in financial services. Can you help give us a sense of the big picture in terms of how AI can be used and is being used in finance? If you look at finances, historically, it has been very technology-driven industry, all the way to ATM to even using digital banking. It's very data-driven industry. Two very fundamental theses that you need for AI transformation is you have to have a lot of data, and it has to be digitized. Those two key ingredients have always existed in finance. The third piece that is very interesting is there's no agency problem that typically exist in other industry. For healthcare, the people who own the data have huge incentive to monetize it. As a result, a decision-making process that is typically involved in any AI transformation is very fast for finance. I think those three combination make finance as an industry very ripe for disruption to AI. There are multiple things that are happening. AI is now being used to disrupt how you interact with the front office, all the way to the back office. I can give you multiple examples of how that is being done. Let's maybe start with the front office, front office meaning how the company's interacting with the customers. Tell us how it's changing, how financial services companies are interacting with customers? In front office, the most important thing is customer interaction. AI is being used to fundamentally transform a omni-channel experience. Now rather just going to online website, you can have a conversation with your bank. That's one area that is being transformed using advances in conversational AI. Better micro-targeting using unique signals that typically banks never used in targeting their customer base. The third is personalization. Personalization has been there for some time, but creating personalized financial products is another area that is happening on the front side. What do you mean by creating personalized financial products? Certainly we've talked about personalization in this course, it's an important theme, not just in financial services but in retail and so many other settings, but it's also an idea that's been around for a while. Tell us a little more about personalizing the product itself. That's very good. Personalization has been there for a long time. Personalization of messaging has been there for long in terms of email, better email targeting, using the right language in your email. Ad targeting. Exactly, ad targeting. But what has happened in the last few years or last few quarters is increasingly you're seeing products being created. One example is thematic portfolio. You can have a creative portfolio, very unique to you based on your interest. I want to invest in sustainable companies, I want to invest in companies which are targeting a particular region. How do you get to that without actually mining their SEC data, their filings, their investing relationship, and then coming up with a particular theme? That's one example of how AI is being used to create thematic portfolio. Interesting. You also mentioned AI is being used in the back office as well. Walk us through how AI is being used in back office settings? AI has been used in finance for a long time, traditionally, back office like fraud and AML, anti-money laundering. What has transform what is happening is some of the newer deep learning techniques are being used to create very complex fraud detection models. The scale at which you can manipulate data and use new data sources, that's what a huge transformation is happening. Same thing on AML. Nowadays, for example, JPMorgan runs around one-third of all transactions that are happening in the world in some way through JPMorgan, and we are using real-time, extremely complex knowledge graphs to create insights and alerts for fraud. That's one example. Other areas, traditional areas in call center rather than your IVR experiences completely we transform using a conversational experience in customer. You can even have a experienced conversation with your bank account in terms of how much bank balance there is in the bank, in your checking account. You can do simple wire transfers now completely conversationally without ever talking to a human being. That's the kind of back-office operations that traditionally had been done. Now are getting transformed through AI. I think one other example I should talk about is contracts. A lot of back-office operation in bank is associated with contract negotiation and maintaining contracts. There's a lot of new exciting stuff happening in terms of machines understanding contract, and then identifying discrepancies, alerting the relevant parties in a contract to take action. Some really great work going on in that area as well. Apoorv as the global head of AI, clearly, both you individually but your organization as a whole is investing a lot in AI. When companies are investing a lot in AI, the question is, what are the things that need to be in place for you to get returns out of it? Because as we have seen, there are many companies that are investing in AI, but not yet seeing the returns. Tell us how you think about this. What is an AI strategy that you find works well and is likely to produce returns in the long run? That's a very good point. I think one thing to note is it's very easy to get started in AI. I think literally you can hire few Data Scientists, give them a laptop and they should be up and running, doing AI or doing machine learning or data science. I think where you're hiding is how do you have huge impact, transformational impact in AI? I think there are three ways we have seen that work across. One is you have to have the right infrastructure to do AI at scale. What does that mean? Essentially, it means having your data in place, data easily discoverable, easily annotated and then ability to train large-scale models. That's essentially what I call AI infrastructure and Facebook, the Googles of the world invested heavily in this and that's the reason they have been able to leave throughout the industry. The second piece is you have to look at your business processes from an end-to-end perspective and see how AI can be applied. End to end process rather than a small piece of the overall process. The third piece I would say, you can use AI and should use AI to create new digital experiences. This is where you will see the most impact over time. What's an example of a new digital experience that you can create through AI? I think AI has been able to do amazing new experiences to conversation. Speech to text accuracy has increased a lot. Ability to create completely AI-generated text has increased a lot. Now you can think of experiences that were just not possible two or three years back, just two or three years back. Given an example, if you're talking in the private wealth space, there are a lot of low touch, high-frequency transactions that you typically as a bank want to automate. Example of, hey, you wake up in the morning and you want to check your portfolio, how is it doing and why is it down? You wake up, talk to Alexa or one other digital experiences assistant and say, "Why is my portfolio down by two percent?" Then the AI system goes and dissects your portfolio and comes up with very good reasons of why it is down and gives you that transaction. Of course, in all this experience, if you want to dig deeper and go very deep, you actually call your wealth advisor, but that's the kind of experiences that we're talking about. It's certainly clear how you can apply AI to improve existing experiences. But when you talk about creating new experiences what exactly do you mean? AI has enabled completely transformational experience and how you converse with machines and how you can generate new text, machine-generated texts. These two advances can be applied to transform how you interact with the bank. Given an example, typically a low touch high-frequency interaction with your private wealth adviser is typically something that bank want to automate. Example of this, is you wake up in the morning, you talk to your bank through Alexa and say, "Hey, how is my portfolio doing today? How did I do yesterday in fact?" It says, "Hey, it's down by two percent." Then you start digging deeper into, "Hey, why is it down by two percent?" It says, "This is how your portfolio is constructed and this is the reason why it is down so on, so forth." That kind of interaction is very low touch, very high-frequency, but something that you easily can automate it today. That's the kind of experience that I'm talking about. Well, there we have it, three pillars that Apoorv is suggesting the first one being, before we embark on big initiatives with AI, let's make sure the underlying infrastructure, the Data Infrastructure and platforms are in place. The second is, don't think about individual touch-points alone, but in fact, think about entire end-to-end processes and look at how AI can transform that entire process. Lastly, use AI certainly to transform existing experiences but also think hard about how AI can create completely new experiences for the customers. Those can be strategic and game-changers for the company. Apoorv, thanks again for joining us. Thank you for having me.