Today, a cellular network can anticipate issues and course-correct before a single outage. Chatbots are helping close trouble tickets by engaging in human-like dialogue with customers. A manufacturer can analyze weather patterns to mitigate risk to its delivery fleets before a drop of rain has fallen or a shipment is missed. And predictions by machines are already exceeding the ability of pathologists in diagnosing some diseases.
AI has been a long time coming, yet almost suddenly it is gaining ground with remarkable speed. Machine learning, the workhorse of AI, can see into blind spots, and spin through pattern analysis that would be too expensive and time-consuming to perform otherwise. For example, Google’s deep neural networks achieve super-human performance in image classification. Applications of machine learning are turning into real, tangible business outcomes. And the pace is accelerating.
Baidu, Google, Amazon and Facebook may have been at the forefront, but implementing AI capabilities across all stages of the product design and development process is fast becoming table-stakes for companies looking to innovate and improve customer experiences. Today, 1,652 AI startups and private companies have raised over $12.24 billion of funding.
Why now? Computation is more powerful than ever, storage is just a cloud away and “data” has become a part of the consumer lexicon. Simultaneously, building with AI has become democratized through APIs. In short, the necessary digital foundations are finally ready for AI.
What problem are you trying to solve?
To incorporate machine learning into products is to enter a world that is exponentially more dynamic and more capable of nuanced variations. And AI will become core to every product and service -- and stimulating new experiences and revenue streams.
There are two primary categories for AI use cases in Product Development:
- Optimize User Engagement: Opportunities to augment and amplify customer experiences through new ways of interfacing and working with machines
- Optimize System Outcomes: Opportunities to arrive at the best available solution to hard problems be that in development, operations or maintenance
Of course, these use cases do not work in a vacuum. Rather, they work in tandem; to predict failures in operations or increase test coverage by 30% is to enhance customer experiences.
Optimize User Engagement
AI is meant to defy convention. Google Home and Alexa are driving a new voice UX into households. The Toyota Concept-i has a personality that manifests as a virtual assistant. The Yumi interaction design is a playful dot that moves around and directs you. In the next few years, companies will routinely customize the design of their products and services to match the needs infinitely different archetypes and in so doing elevate the value of the product in the eyes of the user. However, to avoid the band of abandoned...Form must follow emotion. Design for people. As we move into the world of AI, Immersive and captivating experiences become paramount. Relationships and "emotional" awareness leads to less frustrations and more enjoyment. More be-spoke interactions.
Optimize System Outcomes
AI is proactive and it's directive is to converge on the best possible outcome. ABB claims it gets 55 percent of sales from products that are digitally enabled and is working with IBM Watson software analytics to drive greater uptime, speed and yield for industrial customers. Network operators can define the intent of the network and in simple language and let the network do the rest, selecting a correct set of services from the available catalogs to implement the day the placement of virtual network functions. Or take, Amazon’s Gamelift that continually scans available game servers around the world and matches them against player requests to join games. If low latency game servers are not available, the service's auto-scaling feature can automatically start one, hundreds, or even thousands of game servers across Amazon Web Services global regions. Companies can no longer passively wait for a major crisis, like an energy outage or a product malfunction, which can damage a brand’s reputation, nor can they rely on generic, impersonal user experiences. Making predictions that get better over time is one thing machine learning products do very well.