Artificial intelligence (AI) has moved rapidly from the realm of science fiction to become groundbreaking technology or even just a marketing gimmick from some manufacturers. Consumers, content providers, service providers, and device manufacturers are all still trying to grasp exactly what AI is, how it can impact them, and how they can better use it. At a basic level, it is just the output of some sensory information, and the functions that on-device AI can improve on depend heavily on the sensors available to that device. Ovum has therefore divided AI on consumer devices into categories covering different sensory inputs: voice, vision, movement, and environmental, with different levels to gauge the capabilities of each device along each of these vectors.
Figure 1: Installed base of voice-recognition capability by device
Source: Ovum's Consumer AI Forecast 2017–22
This division should serve to highlight some key facts about AI. Notably, a total for "all consumer AI devices" is a useless metric and will only be used to make marketing sound more impressive. The AI on a smart TV and the AI on a smartwatch will be radically different, and some devices will have very little or no overlap in their capabilities, which means that an "all consumer AI devices" metric is worthless for estimating the potential reach of any service. Instead, grouping AI devices by their capabilities gives us much clearer insight into how the potential markets for specific services will evolve and which devices will act as the lynchpins of the consumer AI experience.
The big issue for AI on many consumer devices will be cost. The price of the specialized chips with neural engines that we see on the latest Apple or Huawei smartphones will inevitably come down in price, and offloading calculations to the cloud provides a more immediate and very cheap means to access the power of AI. However, some devices will still demand on-device or at least local capabilities because of privacy and security demands, and Ovum believes that we will see the rise of the "local cloud" for AI services, with a home hub or smartphone controlling the AI functionality and handling the bulk of the workload for a wide network of cheap smart devices in the home. This future places an increased importance on the current smart speaker battle because the smart speaker is the ideal candidate to serve as this hub.
Straight Talk is a weekly briefing from the desk of the Chief Research Officer. To receive this newsletter by email, please contact us.