{"id":68537,"date":"2018-12-12T17:17:54","date_gmt":"2018-12-12T17:17:54","guid":{"rendered":"https:\/\/news.microsoft.com\/?p=429399"},"modified":"2018-12-12T17:17:54","modified_gmt":"2018-12-12T17:17:54","slug":"arccos-golfs-virtual-caddie-uses-ai-to-improve-your-game","status":"publish","type":"post","link":"https:\/\/sickgaming.net\/blog\/2018\/12\/12\/arccos-golfs-virtual-caddie-uses-ai-to-improve-your-game\/","title":{"rendered":"Arccos Golf\u2019s virtual caddie uses AI to improve your game"},"content":{"rendered":"
<\/div>\n

Some golfers have never known the pleasure of playing with a caddie who understands how far they can hit with each club, who knows the layout of the course or who can predict how the wind and weather will affect their game.<\/p>\n

That\u2019s why Arccos Golf<\/a>\u00a0used tools in Microsoft Azure to create the first virtual caddie\u2014 one that uses artificial intelligence to parse all that data and offer personalized recommendations about what clubs to use and how to play a hole.<\/p>\n

The goal? To create a level playing field for all golfers and help newcomers and longtime golfers alike improve at a quicker pace.<\/p>\n

Jack Brown, Arccos senior vice-president of product and software, chatted with Transform in advance of a recent AI in Business<\/a> event in San Francisco about developing its AI-powered caddie.<\/p>\n

TRANSFORM: How does technology fuel innovation at Arccos?<\/strong><\/p>\n

BROWN: Fundamentally, we are a group of people who love new technology and seeing how we can apply it to golf. We\u2019re always thinking, \u201cHey, there\u2019s a new tool out. Can that help a golfer improve their game?\u201d<\/p>\n

TRANSFORM: How did Arccos get started with AI?<\/strong><\/p>\n

BROWN: We didn\u2019t start out with the mindset of using AI or creating a virtual caddie. The original thought was, \u201cHey, we\u2019ve got this great hardware concept where we can put sensors on every golf club and use them to track the distance that individuals are hitting each of their clubs.\u201d<\/p>\n

So right away we were able provide users with analytics and insights around parts of their game that they\u2019re struggling with and what they need to work on. But then we had a lot of users asking if we could use that data to also recommend what clubs they should be using.<\/p>\n

So we thought, \u201cWhy don\u2019t we use AI to create a virtual caddie that can use all your data and environmental data to offer that same kind of personalized advice as a human caddie?\u201d<\/p>\n

TRANSFORM: How did you move from idea to implementation?<\/strong><\/p>\n

BROWN: We started researching what tools were out there to help us take that next step and actually create this virtual caddie. We knew we needed a scalable tool that could accommodate the complexity and breadth of the data we were processing. So enter Microsoft Azure.<\/p>\n

We\u2019ve been using Microsoft Azure Cosmos DB<\/a>, Azure Machine Learning<\/a> and Azure Kubernetes Service<\/a>, and the thing we love about the system is that it\u2019s just extremely fast, which is really important when a user is out there on the course pinging the AI. You\u2019ve got other golfers in your group and the ones behind you, and no one wants to sit around waiting for your virtual caddie to respond.<\/p>\n

TRANSFORM: What were some of the initial challenges you faced in using AI?<\/strong><\/p>\n

BROWN: Probably the biggest hurdle right off the bat \u2014 because we are so data rich \u2014 was to figure out how to keep things simple. Initially you think, \u201cHey, I want to know how much it rained yesterday because the grass will be wet and the ball won\u2019t roll as much,\u201d and then you have to say, \u201cNo, stop. Don\u2019t grab everything just yet. Let\u2019s do something that works first and really shows value to our customers. Then we can build on that success.\u201d<\/p>\n

Ideally, we wanted our virtual caddie to be able to give you a recommendation regardless of where you are on the course, which is what we have today. But in our initial version we decided not to use any environmental data and just use the shot data from Arccos users and only give you a recommendation on how to play a hole from the tee. So if you slice your shot and you\u2019re now over in the rough, you couldn\u2019t follow that strategy anymore. But it was a good starting point, and users thought it was awesome.<\/p>\n

The next update took environmental data like wind and elevation into account. Then Arccos Caddie 2.0, which we released with Microsoft earlier this year, is now able to recalibrate its recommendations on the fly and give you a new strategy regardless of where you are on the hole.<\/p>\n

TRANSFORM: Have the benefits from AI accrued over time?<\/strong><\/p>\n

BROWN: You do get more transformative benefits the longer you use AI. When our virtual caddie is deciding what club to recommend, it\u2019s looking at your data, and data of every other user like you. By now we\u2019ve captured over 100 million shots from our users, we\u2019ve mapped 40,000 golf courses and we have more than 1 billion geotagged data points on those greens.<\/p>\n

And as we get more and more players using the Arccos ecosystem and playing on the course you\u2019re playing on right now, our recommendations continue to get better and better, and that would be the case even if we didn\u2019t continue to improve our models. And the proof is in the pudding;\u00a0Arccos users are improving by 3.79 strokes in their first year of\u00a0using our virtual caddie.<\/p>\n

TRANSFORM: What advice would you offer to companies just getting started in AI?<\/strong><\/p>\n

BROWN: Really think through whether your use of AI is a novelty or something that\u2019s truly beneficial to your users. Some people are using AI in ways that aren\u2019t really adding much value. It\u2019s really something neat that allows you to say, \u201cHey, I\u2019m using AI\u201d but you probably could have just used a complex algorithm.<\/p>\n

Top photo:\u00a0Arccos Golf SVP Jack Brown demonstrates Arccos Caddie at Conversations on AI, a Microsoft event in San Francisco.<\/em><\/p>\n

Related:<\/p>\n