Google has generated some buzz over the past two years for their study of Artificial Intelligence (AI), and the potential ways that AI could be used in the future. CSRA’s CTO Seth Abrams attended a Google Cloud Next Conference in Washington, D.C., where he gained insight as to how Google’s AI development program could be of use to CSRA and its customers. We sat down with Seth to hear his key takeaways from the conference.
Thinking Next: What were the highlights of the Google Cloud Conference?
Seth Abrams: The highlight was the introduction of Google’s Artificial Intelligence platform to the government community. They gave a great example of how AI can be used to solve almost impossible problems.
Google has focused on a game called Go that is extremely popular in East Asia, where you don’t know who the winner is going to be until the last moments. The game is so complex, it’s hard at times to gauge who is winning. So they interviewed players asking “Who do you think is winning?” to try to understand the nuances of the game.
Google wanted to prove that their analytics engine can learn intuition to predict who is winning. They developed a neuro-computing model that would play against itself to learn better ways of playing and determine what put someone in the lead or not.
TN: How were they able to accomplish that?
Seth Abrams: Google realized that conventional logic doesn’t apply to problems of this complexity, so by leveraging an AI approach that focuses on a trial and reward system, they can solve problems through rapid iterations of the problem. Each iteration leads to a reward system that gets closer and closer to solving the problem, much like the way humans learn.
Google realized that if they’re going to do this kind of computing, they’re going to run out of computation power because it’s so intense. So they built their own chip and data centers for neuro computing.
TN: How is learning Google’s point of view on the journey to cloud going to help CSRA and our customers?
Seth Abrams: During a session I attended about cloud migration, data gravity was described in the most interesting way: your data lives is where your cost is. That’s extremely important to think about when making architecture decisions.
Many cloud migrations get stalled when they don’t think about their data first. It’s always been CSRA’s strategy to look at the entire mission space, but considering data gravity first would be a new approach for us—and it makes sense to switch over.
TN: How are you going to implement the information you learned?
Seth Abrams: Looking at our customers’ mission and the challenges we face with them, we’re trying to find an algorithm to fix the problem. We have situations where that doesn’t always work, or it’s very hard to manage, because what ends up happening is your conditions grow beyond what’s manageable. Artificial Intelligence can understand those complicated models and be able to somewhat react to them, which then lessens the cost and increases the validity of our services.
What we’ve done with predictive analytics is similar, where we can now anticipate problems before they happen and remediate them with AI. We can then design better applications and reduce the amount of care they need to stay in production, because they can learn as opposed to just staying the same, and can realize “I got 17 error messages doing the same thing; maybe I shouldn’t do it that way.” It will start to cut out the what if scenarios and turn it into an active problem set.
Google has set themselves apart from the crowd, as this multi-national, multi-billion dollar company that is betting the farm on AI. CSRA will definitely be watching to see what they do with this strategic path.