Artificial intelligence is becoming a central concept in modern enterprise applications, with greater capability driving the development of more use cases. As AI technologies develop, vendors must continually reevaluate where they are most and least appropriate.
AI is here, but where do we want it?
The study of artificial intelligence is not new, but it is only in the past five to 10 years that computing technology and scientific development have yielded something in marketable form. Programmers are continuing to refine neural nets, algorithms, and deep learning systems, feeding increasing amounts and types of data into the system, resulting in richer behavior that is applicable in a growing number of situations. Machines that can pass the Turing test are no longer science fiction or even particularly novel; businesses use conversational AI every day to shoulder some of the load of customer service and e-commerce tasks that would otherwise require human labor, and consumers have access to AI assistance through products like Alexa and Siri.
If modern AI can do so many things now and is continuing to improve, then it is reasonable to assume that it will be able to do more as time passes. What businesses and other implementers must do is consider how and where to put this tool to the best use. Poor decisions regarding AI could damage business performance and brand image. Microsoft's unfortunate experience with the Tay chatbot AI, which learned to behave literally like a racist, conspiracy-theorist teenager, should be instructive of this.
Ovum believes the primary areas of AI utility are machine-to-machine management and fault detection, inbound customer service, and business intelligence. The Internet of Things is built in part on the idea that machines and sensors can communicate directly, optimizing functions and warning of potential equipment failure. Chatbots have already proved themselves capable of fielding mundane help requests by leveraging knowledge bases and can also perform some transactions; NLP voice systems can and have enabled this for phone service, notably among banks and utilities. In terms of BI, many vendors include such functions as trend analysis, anomaly detection, and predictive analytics.
These are not the only places where advanced AI can provide business benefit. Sales applications may include automatic follow-up, next best action, deal prediction, and several others. Marketers use AI to optimize and manage new campaigns, test variants, and spot trends in their customer base and their industry. However, Ovum believes that outbound uses of AI should remain limited to mundane alerts and drip contacts for the present. While it is certainly possible to create an application that is proactive in customer outreach, its ability to predict individual behaviors would be disappointing, and there is also the "creepiness factor" of having a computer call a consumer to offer a product or service based on data the individual didn't even know had been gathered.
Artificial Intelligence Technology and Applications: Ovum Definitions and Taxonomy, INT002-000069 (February 2018)
"Ovum introduces artificial intelligence reality (AIR) chart," IT0014-003311 (July 2017)
"Artificial intelligence and machine learning are not the same," IT0020-000299 (July 2017)
"Five ways customer service can make the most of artificial intelligence," IT0020-000272 (April 2017)
Marshall Lager, Senior Analyst, Customer Engagement