ai.bythebay.io: Chris Moody Interview
Recording: ai.bythebay.io: Chris Moody Interview
you [Music] you [Music] it's a good question i think and i don't think i have a great answer for it i think the definition of a i just varies tremendously i think some people think it's just the same thing as machine learning think some folks think that it's logic i think some folks think that it's reinforcement learning all there are different approaches and I don't really know which definition is the right one what I do know is that i want AI to help me learn about the world time so as soon as i have that are tools that are instruments and ultimately leading up to like almost partners and trying to understand the world i think whatever fulfilled those goals whatever helps me be a better scientist i'm ok calling that a icai being successfully used in in totally humdrum lives honestly in in optimizing lots of different problems in finding tiny little patterns and trying to understand the systems that you live in how is that going to change in the future it's extremely difficult to say right like I think it's very difficult to chart the trajectory of something that's science in nature I think scientists across all disciplines have extremely difficult time conducting what the future is going to look like and now if things like AI it's also businesses are jumping into the same for a trying to have to exactly same questions how can i prepare how should I devote my resources and how should I think about how to devoting it hey I'm I really don't know how it's going to change in the future you can see the hardware certainly changing a lot you can see that like the software is changing a lot you can see that I think actually probably not i think about actually probably the biggest trend is actually seeing data fall into siloed area so you can start to see that data is a moat that protects a company and its its competitive advantage and that's very different than say a place where you have data that's free openly available that data is what pushes new algorithms but when that data only exists inside of one company saints facebook or the googles of the world only they get the resources to get the best AI i can definitely excrete that as a trend going forward especially with things like clouds being generally isolated services are isolated to the rest of the world where data becomes very very particular to the only the person who owns it and not available to anyone else I see that as more of a worrying trend inside of interpretability right and this is more or less what what worries me what keeps me up and it ties into I think one of the very one of the first questions you asked me like what do I actually define is AI and to me it's something that helps me do science better that means that it can't be a complete black box system it can't just tell me what's good and what's bad I have to actually be able to take that model apart understand the different parts and components and that helps me understand the system that's going on underneath and helps leverage whatever inside I might learn about my company my business whatever the underlying physical system might be right you don't see a lot of machine learning for discovering new phenomena and I think that could be something and I think that's because so many systems are not been interpreted if you have that focus on that interpretability then you have a partner in doing science nothing you can sir the people it's incredible to come in here and see the folks that you know are doing amazing work everywhere else and to see what they think is incredibly important where they are focusing their time see what kinds of learnings you can leverage from them it's that's an incredible opportunity of a mountainside [Music] you