While the media frets about computer superintelligences potentially usurping the human race, the general public is just as concerned about a more immediate issue related to artificial intelligence (AI): the prospect that AI will have a negative impact on personal data and privacy.
To alleviate this concern, makers of tablets, smartphones and smart speakers are developing products that use the capabilities of 5G to perform visual AI processing tasks by edge servers and appliances, bypassing the privacy risks involved in sending data to the cloud. By 2025, two out of three smartphones are expected to include built-in AI capabilities. Global revenues for AI smartphones are forecast to increase to $378 billion, up from $29 billion in 2017.
In a survey conducted by Omdia in 2019, 29 percent of respondents indicated that they are uncomfortable with the idea that AI will soon become smarter than they are. This is same percentage that expressed discomfort with AI’s impact on privacy.
Visual AI applications raise particular privacy concerns, as consumers are increasingly worried about what images and data are is being captured and shared with third parties and how much of this information is being used to train AI models.
Deep-learning object detection, image and face recognition are the primary drivers for vision-based AI applications. In 2020, expect to see an acceleration of this trend, as the visual AI software and hardware stack becomes commoditized and scalable. In the consumer domain, embedded vision-related AI will be most prominent in tablets, smartphones and smart-home speakers.
Today, visual AI processing is mostly implemented on-device at the edge. However, 5G allows for AI inference processing to be offloaded from the device to an edge server or an appliance. The recent announcement by Qualcomm of a 5G-enabled intelligent edge appliance being built for a certain hyperscaler is proof that AI applications and 5G go hand in hand.
At MWC20, NVIDIA will be beating the AI and 5G drum in relation to its EGX platform announced last year at MWC Americas. There is a chance that other mobile infrastructure vendors and CSPs will chime in as well. Edge processing has benefits over cloud processing when it comes to latency, bandwidth, cost and privacy.
On-device AI processing is the best antidote to privacy concerns, with Apple using on-device AI to dispel any privacy questions for the iPhone. However, edge server or edge appliance processing is a grey area with lack of clarity on questions including who owns these edge appliances, whether there is any data being passed over the cloud and what must be done to ensure privacy protection.