Connect with us


Mobile gaming is putting machine learning right in your pocket- here’s how

Will machine learning start powering your favorite mobile games?

person on phone playing a machine learning game near train tracks
Image: Unsplash

The global gaming market is expected to top $180 billion in revenue by 2021—good for 31 percent growth from last year’s level. Over half of the market’s revenue comes from mobile games, which also accounted for 74 percent of all consumer spend in app stores in 2018.

Meanwhile, the newer machine learning market is a bit smaller, though with even bigger growth: expected annual growth north of 40 percent should bring its revenue to $23 billion by 2023. But increasingly, these markets are anything but separate. Companies like Playgendary, one of the world’s largest mobile game publishers, has been using machine learning since the beginning. Its games, including Kick the Buddy, Polysphere, Bowmasters, and Tank Stars, have attracted over 500 million users.

But first, the basics: Machine learning is one form of artificial intelligence

In a nutshell, the machine receives, analyzes, and is trained on a large amount of data. For instance, let’s say you want to teach a machine to distinguish a corgi dog from other breeds. You need to “feed” the machine many pictures of corgis in various situations—close-ups, photos in the dark, photos of the dog half-turned, and so on—so that the machine understands what features distinguish it. This technology has many applications in the gaming industry, from analyzing the levels of a game to creating smart characters that adapt to a user’s playing style to personalizing advertising.

Through machine learning, for example, Playgendary tracks when a large number of players cannot pass a certain level—a sign that the level is too complicated or contains a mistake—and makes adjustments accordingly.

Relatedly, by analyzing historical data, Playgendary noticed user patterns in players who left the game, like the fact that 3% left after 20 unsuccessful attempts on a level. One possible solution: reduce the complexity of the level on the 19th attempt to avoid frustration and subsequent quitting. Similarly, when a player begins using the game less and less, the algorithm notices that trend and offers some kind of bonus (an in-game asset, in-game currency, some exclusive content, and so on).

To that end, in-game assets can also be targeted to users—both to fit their playing style and their budgets

Say, for example, a player is using a two-handed sword but is using it ineffectively. Machine learning can identify that behavior and offer them a better weapon to fit that style—possibly a one-handed sword or a bow.

Additionally, many games have millions of players; those players obviously don’t all have the same income. The machine learning algorithm can determine and offer the best item at the best price for the player based on the data that the game has collected about them.

This approach has been shown to grow in-game purchases by as much as 40 percent. Targeted offers are especially important for overcoming the psychological barrier of the first purchase. By making a unique offer at a comfortable price at the right time, it’s easier to gently push a user to buy something for the first time. Once the player realizes those purchases are actually improving and personalizing the gaming experience, more are likely to follow.

Similarly, machine learning can help widen the audience of a game by determining the common characteristics of its most active, high-spending users and then importing that data to an advertising service to find similar ones. Monetization is crucial because, at the end of the day, businesses need to make money and cover their costs. But machine learning’s potential is greatest with regard to how it will actually shape gameplay.

In the future, completely personalized games could be possible

With avatars that more closely reflect their owners, both in appearance and psychology. Machine learning will allow developers to create smart games that adapt to the user’s playing style, while non-player characters will become smarter and actually reflect the context and history of a particular player’s interactions. To that end, dialogues between characters will also become deeper, taking the gaming experience to a new emotional level.

All in all, machine learning and mobile gaming are — as our opening data points suggested — just getting started. But as both markets gain momentum, it’s safe to say they will increasingly overlap—putting machine learning right in gamers’ pockets.

What do you think? Are you interested in seeing more machine learning in your video games? Let us know down below in the comments or carry the discussion over to our Twitter or Facebook.

Editors’ Recommendations:

Click to comment

You must be logged in to post a comment Login

Leave a Reply

More in #GameTechie