Interview with Javier Luraschi (Software Engineer, RStudio) - Speaker at Global Artificial Intelligence Conference April 2018 Posted on : Apr 20 - 2018

We feature speakers at Global Artificial Intelligence Conference - April 27 - 29 2018 – Seattle to catch up and find out what he or she is working on now and what's coming next. This week we're talking to Javier Luraschi (Software Engineer, RStudio) Topic - "Modern Data Science With R"

Interview with Javier Luraschi

1. Tell us about yourself and your background.

I love to building software products. I've been building them since I was 9, then I went to college to study Math and Software Engineering, worked at SAP, Microsoft, Microsoft Research and RStudio, so far so good.

2.  What have you been working on recently?

I've been focusing on making R work smoothly with big data and deep learning.

3. Tell me about the right tool you used recently to solve customer problem?

This happens to work the other way around for me: Users come with particular problems that they expect the tool should solve and I help them out. Recently, around improving support for time series and performance improvements while running R code at scale.

4. Where are we now today in terms of the state of artificial intelligence, and where do you think we’ll go over the next five years?

I see the first wave of AI already going mainstream, as in, everyone can have their own classifier and use deep learning in their projects without having to be an expert in the field. On the other hand, active AI research is going beyond classifiers and focusing more on agent systems that can actually take actions for themselves and learn from them, the progress of AlphaGo Zero is the highlight of 2017. Five years from today, I think we are going to see AlphaGo-like systems going mainstream, for instance, it's going to be really easy to create a Snapchat bot that creates interesting content and learns from feedback entirely on its own and those agent-centric AI algorithms are going to have applications in driverless cars, search, etc. On the research side, I think we are going to find ourselves asking AI to start self-improving itself, some of that is already happening with Google's AutoML but it's just the beginning, I think 5 years from now, researchers are going to have a hard time catching up with the learning models the learning systems themselves are discovering. Some complain today that the deep learning model parameters are not understandable by humans, in 5 years, we are going to complain that the models themselves are not understandable by humans and yet, produce meaningful predictions and behaviors that we will use without hesitation.

5. There is a negative perception around AI and even some leading technology folks have come out against it or saying that it’s actually potentially harmful to society. Where are you coming down on those discussions? How do you explain this in a way that maybe has a more positive beneficial impact for society?

Right, well, there are definitely arguments to be a bit anxious about this, I mean, things are moving pretty fast and, like any disruptive technology, it always can be used for good or bad. Think of nuclear physics, on one hand, you have weapons of mass destruction and on the other one, you have nuclear plants. I do believe that the world is mostly filled with good people so I'm very optimistic about that. If we can create a system that can solve generic problems, it really opens up the door to help us solve a lot of relevant problems. I think this is especially interesting in robotics with applications to warehouse packing, food production, driverless cars, package delivery, etc. Those tasks seem pretty trivial to us as humans, but robots can help us automate all these tasks and with the right government support and incentives, get society to learn new skills and raise the standard of living.

6. When you’re hiring, what types of people are you hiring? The job market for traditional programmers, engineers is very difficult to get into AI space. Are you hiring from that talent pool or is that a different talent pool? In terms of talent, how do you go about ensuring you get the best AI people at your company?

We don't do a lot of hiring for AI jobs per se, but I'm pretty sure that AI is going to be relevant for everyone and I'm also sure the AI skills we need today are not going to be the same in two years. Therefore, I believe it makes more sense to hire people that are self-learners, self-motivated and aligned with what you are trying to build, if we do that, I think we are set for success.

7. Will progress in AI and robotics take away the majority of jobs currently done by humans? Which jobs are most at risk?

Not take away but definitely be part of the majority of jobs done by humans, if you think about web search, it's already part of the majority of jobs: Doctors search for new treatments, school teachers for new teaching content, marketers for better ad-targeting, etc. it's everywhere, but we still have all those professions, I think something similar is going to happen with AI. Doctors will have cameras and sensors that automatically diagnose patients, school teachers will get delivered content that works for their classroom, marketing will get customized ad-campaigns that automatically picks the content, schedule, etc. I think people are going to love this interaction and is going to make the world much more efficient and will free up professionals to make progress in other areas. I would say that transportation jobs are the most at risk. Transportation is a significant percent of the economy and there is room to create a convoy of trucks where one single operator can manage ten trucks, but it's also an opportunity for them since operating high-tech convoys it's going to require a different set of skills and is also going to happen gradually; therefore, I'm really hopeful that this won't be disruptive but rather an opportunity of growth for everyone, I'm an optimist.

8. What can AI systems do now?

I think they are really good at classification problems, throw them any data set: bank transactions, website clicks, store purchases, etc. and they can create models that are really good at predicting where an element belongs to. It definitely has important applications already, for instance: is this bank transaction fraudulent? Should I show page A or B from a given a set of clicks? Should I offer a discount on some product given this purchases? etc.

9. When will AI systems become more intelligent than people?

They already are, in many fields. Boardgames is an obvious one, video games as well, image classification, radiography, etc. AI systems are better than the average person, but not better than the expert in many cases. But maybe the question here is as to when will the same AI system become more intelligent than people across many tasks? As in, solving Sudoku, playing music, playing chess and having a conversation. There is a lot of research being done today in transfer-learning and general-problem-solvers, I don't think we are more than 5 years away from having an AI system more intelligent than the average person. It's a close race but we don't realize this because we are heavily biased towards humanity maintaining the lead. If someone from the 1900's could travel in time and take a look at where we are today, I'm sure their perception would be much more different, we get used to technological advances really fast. I remember that Hacker News went crazy when SpaceX landed their Falcon rocket for the first time, next landing everyone was like, no big deal.

10. You’ve already hired Y number of  people approximately. What would be your pitch to folks out there to join your Organization? Why does your organization matter in the world?

Well, the first step would be to check out rstudio.com or attend my talk. As for relevance in the world, I believe it's huge. We build tools that are used by millions of users and many researchers, say, epidemiologists, biologists, sociologists, economist, data journalists, data scientists, etc. so the tools we build help make scientific research a bit easier, a bit more productive and with that, help accelerate advances across many industries which have a real positive impact in the world.

11. What are some of the best takeaways that the attendees can have from your talk?

I love hands-on demos so, as an attendee, you will get a first-row-seat to see how R and RStudio can help you get started in modern data science or help you be even more productive as a proficient data scientist.

12. What are the top 5 AI Use cases in enterprises?

5 or 50 use cases? I honestly don't know, it's hard to pick overall themes across industries, but they tend to boil down to increasing revenue or reducing costs. For instance, if you are running ads, AI can help you find ways to increase click-through and conversion rates to increase sales or, it can also help you make your ad-campaigns more targetted to help reduce costs. I'm curious to see how other speakers tackle this question.

13. Which company do you think is winning the global AI race?

As in 2017, I would say Google/Alphabet finished first, in particular, DeepMind which operates as a subsidiary of Alphabet, I believe.

14. Any closing remarks

I'm looking forward to meeting some of you at this upcoming event, thanks to the organizers for putting this event together and for this interview, it was a lot of fun.