Steps to incorporate artificial intelligence in app development
To call the marketplace for new apps crowded is a sizeable understatement.
As of the first quarter of this year, there are 2 million apps on Apple’s App Store and 3.8 million on Google Play. From a user perspective, this app abundance is a huge benefit: If one app doesn’t meet expectations, there are untold others that can.
Yet from an app developer standpoint, the sheer number of apps available creates an enormous challenge to building a consistent user base. For developers, the strategic implementation of artificial intelligence can help them refine their app in a way that optimizes usability and sets them apart from the competition.
But to effectively incorporate AI in app development, developers need to look at it as more than a buzzword. Instead, they need to ask substantive questions about how AI can specifically and measurably improve the product they’ve built – and then take calculated steps to make that happen.
Looking beyond “powered by AI”
Among app developers, discussions about incorporating AI can quickly become anxiety-fuelled sessions about keeping up with the Joneses. If apps B and C just announced that they’re now “powered by AI,” doesn’t it logically fall on us, the developers of App A, to do the same?
But anxiety and the perceived need to keep pace with competitors won’t create a solid foundation for effective use of AI. Because the reality is that “powered by AI” is a broad term that encompasses many different solutions, and this lack of specificity can and does create confusion about how app developers should regard and incorporate AI.
As a result of this confusion, some development teams use the term more as a PR tactic than an actual indicator of strategically-deployed AI. The vague and marketing-driven use of “AI” can detract from its concrete and useful applications, which include:
- Shopping recommendations: Amazon offers an instructive example of well-deployed recommendation-focused AI. Given that it has a vast product catalogue that far exceeds anyone’s purchase history, Amazon’s internal team designed a neural network that can make the best possible recommendations when there’s hardly any training data. Because its calculations require neural networks with hundreds of thousands of nodes, Amazon developed a way to spread the work across multiple processors in a single server.
- Image recognition: If Amazon is leading the charge on intelligence-driven shopping recommendations, Facebook is the trailblazer on the image recognition front. The social media giant uses a nine-layer deep neural network with more than 120 million parameters to automatically tag people in an uploaded photo. The network was trained using a dataset of more than four million facial images, and has a 97 percent accuracy rate.
- Streaming video: No practical app-based AI deployment list would be complete without Netflix. If you’re consistently surprised by the degree to which the streaming site knows your taste — well, there’s a sophisticated intelligent process underriding that. The streaming site uses filtering algorithms to personalize its service, basing recommendations on behavioral data from 84 million-plus subscribers and metadata on more than 125 million hours of TV shows and movies per day. The company also applies machine learning to data to identify possible server problems early and prevent service interruptions.
Of course, the common denominator among Amazon, Facebook and Netflix is that they’re all massive companies with no shortage of time, talent and resources to channel into AI-driven innovations. But how can app developers operating at a smaller scale also adopt AI solutions?
Steps to incorporate AI in app development
Once developers move past the feeling of urgency to label their app as being “powered by AI,” they can begin to explore what AI deployments could actually mean in practice. And when it comes to substantive applications of AI-driven technology, the list goes on: From optimizing user insights to identifying weak points in app design, AI enables a more personalized, continuous and engaged app experience.
As app developers begin to explore the unrealized potential of AI, they’ll all confront the same question: Where do we begin? Here are two of my recommendations:
- Understand your product’s value prop: AI isn’t going to create value for your app; it will augment the value that already exists. As a first step toward strategic AI implementation, app developers should work to understand what your unique value proposition is. Try to identify where and how you are saving your customers time and money. Are there places where things could potentially be more automated so a human doesn’t have to be involved?
- Focus on relevant content: Most apps fail to offer relevant content that continuously engages their users. As a result, 77 percent of users never use an app again 72 hours after installing; 90 percent of users stop using the app after a month; while by the 90-day mark, only five percent of the initial user base is still actively engaged. In short, the odds are against long-term app use. Given users’ penchant for ephemeral use, app developers can benefit significantly by channeling AI efforts toward strategies and tools for more continuous engagement.
After primarily being a tool of large enterprises, AI is becoming more and more mainstream and accessible to any developer. Through subfields that include machine learning, predictive analysis, and deep learning, AI has helped businesses that are on mobile take personalization to the next level, both in terms of the features that they offer and the marketing efforts that they follow.
As a developer attempting to differentiate your product among a sea of millions of apps, it pays to use AI strategically. But that means looking past the hype — and the pressure to brand your product as AI — and thinking about how AI solutions can actually be applied to create a more engaging and sustained user experience.