Audioburst Search

Building a Company Culture for Success with AI - With Marsal Gavalda of Square

Automatic TRANSCRIPT

So Marsala I know. I want to dive into sort the Competitive Advantage that can be leveraged with artificial intelligence. But from what you've told me your story of getting into a in the first place is kind of interesting so we'll step a little bit off the normal scripted and get a sense of how you got into this world. Yes things that were coming on. The podcast so Yeah so I. I come from Barcelona Catalonia. Spain right and growing up there develop an interested in languages so languages is one aspect that kind of is the part of the journey of how I ended up being a practitioner of machine learning. And if YOU KNOW ANYTHING ABOUT CATALONIA. Catalan and their Spanish growing up. They're very curious about languages. You know how half of my classmates speak Catalan. The other half Spanish one. If you look at the history as the Roman Empire extended throughout Europe there. Were all supposed to be speaking. Latin right and yet it gave rise to Italian one place Romanian French Portuguese Spaniel these languages. Don't call so it's super interesting to see what is common across languages. For example languages have nouns verbs straight. These is how we humans perceive the roles of static objects and actions that we do but then the way in which languages put these building blocks together. That's more arbitrary right in Japanese. You put the verb at the end. So you'd say no. The girl a boo greets which incidentally is also how Yoda talks. Then in Japan you also drive on the left right. Certain things are societal conventions where whereas others are true human universal so anyway languages. He's faceting topic and then the other threat is Sort of computers technology. One day that showed up with an h eighty-five on the early home computers with a with a monochrome. Crt Screen Building Keyboard magnetic cartridge for storage thermal printer and then some manuals that really well designed and taught you how to program. So that's what got me cooking to computer science. And that's where I ended up starting as an Undergrad but they don't realize that there was these these nascent field. At the time called computational linguistics now better known as language technologies. Which is the idea that you can actually use computers to do to to process language. I'd speech Recognition Machine Translation Information Retrieval and. I joined that field right at the time that he was undergoing his huge transformation from being a rule. Dreamin sort of top down approach. How do you pars an English sentence while it has a a a noun phrase and over praise and you try to seize their an adjective. They're also how you do speech recognition. You look at a spectrogram and see that you know before close. If there's there's there's no energy and have some rules that say well frequencies are in these range than is probably a free Qatif because it's mostly white noise like for right the literally. That's how the early speech recognition systems worth of having these rules about the the four men of the vowels and how these gets represented when you look at the Spectral DOMAIN WELL. All of that was being replaced by a much more bottom up data driven approach right which is what machine learning is about and so this was really the how. I got into this area because the field of language technologies was undergoing these huge transformation from rule-based to did a data driven right and then so professionally I I work in this area of speech analytics And then also conversationally. I until I finally joined a square where we use 'em not just for natural language processing but also to understand what's happening within our platform all the payments rolled loggins cetera. I was Gonna say I mean Payment. Data's a little bit different than going about like hardcore. Nlp MODERN NLP approaches. But it sounds like obviously there's transferable skills there and you can kind of faster heels and not only that but you can also imagine that. I'm you know I'll timidly a sale is kind of a conversation between a seller and a buyer and so there's also a lot of human interaction and human elements that go into that. That's I don't think I've literally ever heard somebody say that. Like a sale is kind of like a conversation. It's like what like for E-commerce I just don't take literally anybody thinks about that. But if you can you can break it down that way in your mind and make it work. Then that's awesome. So that's how I was convinced. Join Square yeah cool. It must have worked whenever they had said so. Any who now. Now that you're there you obviously have a hard grounding in the science on you're seeing a lot of what's happening at the cutting edge of AI. Obviously squares a pretty renown firm in terms of of You know being being in the high tech domain payments is a fast moving space The theme of of this month's kind of topic is around the competitive advantage of artificial intelligence. Where can ai? Help companies pull away from their competition when you think about how a enables companies to do that. How do you like to frame it? What are the aspects of AI? Maybe the ways of applying it or or types of applications for all. I know that for you really are going to separate winners losers in a bigger way yes some. I think the best way to go about this is to start with some examples. Concrete examples from square. And then maybe we'll generalized between the two larger. Let's go ahead so in the case of We saw from the very beginning that automation and machine learning is key if you look at the sort of the landscape of credit card acceptance ten years ago which is when we started it was a very very manual process. In fact you as a small business would have to file all these paperwork and then got approved. Which was you know? A small percentage you would get these these clunky hardware for For credit card processing not only that but also the business model was going backwards. Because you would have this monthly fee regardless of whether you had any sale right so when we started developing what we're still best known for the little white reader That accepts credit cards by just connecting to your smartphone. Even though you know within these intervening years we have excuse. Really expanded the amount of product surface or the the products and services that we offer to manage small businesses from inventory management to not management with payroll taxes etc. But even starting with the little reader. There was such a huge change because then the entire experience of on boarding a was so so much much easier and then we also had obviously in order to be able to support that to have such an open platform. Were almost anybody can sign up and become a merchant and start accepting payments was also because from the very beginning. We had the mechanisms in place to watch out for bad actors. Right anytime that you deal with with money and so we had in a sense. The advantage of being kind of a tech company from the very beginning and therefore we have already sort of a mindset of data driven and machine learning which basically means that We have literally hundreds of models that analyze sort of all these actions like logging or a payment and we can flag any any sort of anomalies unsuspicious activity

Coming up next

Building a Company Culture for Success with AI - With Marsal Gavalda of Square

Artificial Intelligence in Industry 2 months ago

AI and the Future of Procurement and Logistics - With Edmund Zagorin of BidOps

Artificial Intelligence in Industry 1 d ago

Smart Talks with IBM: The Impact of Watson AI in a Pandemic

Stuff To Blow Your Mind 1 d ago

Enterprise AI Readiness, and Moving Beyond the Hype Cycle - with Stephen Pawlowski of Micron

Artificial Intelligence in Industry 6 d ago

Enterprise AI Readiness, and Moving Beyond the Hype Cycle - with Stephen Pawlowski of Micron

Artificial Intelligence in Industry 6 d ago