Andrew Moore, Head of Google Cloud AI & Industry Solutions

January 25, 2021

Google Cloud AI leader, Andrew Moore shares what he's most excited for in Cloud AI in 2021 and some advice for businesses getting started in AI.

Video Transcript

Speaker: Andrew Moore

Please introduce yourself and tell us about your role at Google Cloud AI.

Andrew Moore: Hello, everyone. My name is Andrew Moore, and I'm in charge of Cloud AI and Industry Solutions here at Google Cloud.

What are you most excited for in Google Cloud AI in 2021?

Andrew Moore: What I'm most excited about for 2021 is the velocity of artificial intelligence. We have invested heavily in the tool chains that allow us to develop new AIs or new classes of AIs faster and faster. And starting in the second half of last year, we've really seen the phenomenon happening where what used to be a big project is a medium project. What used to be a medium project is a small project. As we're able to create these AIs faster and faster. So at the end of 2021 I think we're gonna look back and say theme of this year. was velocity and the fact that making big fancy AIs has become more and more of a rapid fire type of activity.

What one piece of advice do you have for businesses getting started with AI?

Andrew Moore: Okay, here's what I always tell someone who's beginning an AI projects or being put in charge of using AI for some purpose. Your project has to be in three stages, and the 2nd and 3rd stages have to happen together. So let me explain. This first stage is never asking the question. How do I use AI to dot, dot, dot. The first stage is always: what do I really need? What do my customers really need? What is the thing which will make, my service or product just completely, completely different from how it is today? Then work backwards to figure out how artificial intelligence can help get you to that dream. Now, then it's time to run the project. Two parts there, one, working with the data and exploring what's predictive, how effective optimizers will be, how accurate and fair your classifies will be along with the longer term stuff of. Well, even if I can predict, even if I can model accurately, what's gonna happen next on, decide based on it. How do I put that in production and run it. The key in the what's possible data science stage and the MLOps engineering stage is you've got to work on both of those together. Do not start doing the data science without thinking about how it's going to be implemented. That is the big lesson that I personally have learned throughout my career and is one of the ways that at Google we make sure that an AI concept actually gets delivered rather than just sitting as a fun experiment.

Produced with Vocal Video