Gartner predicts that by 2022, 80 percent of smartphones on the market will have on-device AI, and that’s up from just 10 percent in 2017. Machine learning and data processing in the cloud won’t go away, but on-device AI is what is making connected devices, including automobiles, HD cameras, smartphones, wearables, and other IoT devices, smarter and faster.
With on-device AI, voice assistants are more intelligent and useful, and cars are safer without the microsecond lag when connecting to the cloud. Security is enhanced, robotics can take innovative leaps, and health care solutions — and outcomes — are improved. When your AI power actually rests in your hand, reliability becomes a super-power, because it no longer depends on network availability or bandwidth.
The shift from the cloud
In cloud-based AI, information is gathered and sent from a device, for instance a request for translation, to the cloud, which has the computing power to apply the complex machine-learning algorithms that enable those kinds of capabilities.
But cloud-powered AI has always had some disadvantages; since data is sent off-device, issues like latency, or the time between data sent and query answered, are serious in connected car applications, and create inconveniences in consumer applications like language translation. Since you’re depending on an outside network, your AI applications are only as reliable as your connection. On-device AI is also an essential evolution in a world where demand for privacy and security from consumers is growing exponentially. When sensitive data stays in your device, such as voice ID and face scans, it’s not vulnerable to being compromised in the cloud.
Localized AI has become possible because of increases in the computing power of devices coupled with the increasing sophistication and speed of AI algorithms. And consumers are hungry for the kinds of applications that on-device AI is enabling.
The competitive advantage
The way AI has become the competitive differentiator for companies that want to not just stay ahead in their industry but disrupt it, on-device AI is rapidly offering an advantage for companies that want to hook consumers on the advanced capabilities the technology offers.
About two-thirds of smartphone users are picking up their devices every 30 minutes; over 20 percent are on their phones every five minutes. Consumers already rely heavily on AI applications that have become essential daily tools, from virtual assistants like Alexa and Siri to the traffic prediction and travel planning power of Google Maps.
And now on-device AI means that smartphone assistants are getting brighter all the time, with contextual conversations, enhanced noise suppression, and instantaneous, on-the-fly language translation. With video and image social currency on platforms like Twitter and Instagram, consumers are attracted to the leveled-up smartphone cameras with sophisticated on-device computer vision applications that allow them to do much more, and users are upgrading their devices just to get their hands on these applications.
On-device AI is also enabling the connected car of consumer dreams, as vehicle-to-vehicle and vehicle-to-infrastructure technology loses the lag, making road condition corrections in a split second, improving traffic routing, and ensuring the signal is never lost in tunnels or parking garages.
In the consumer robotics arena, devices with built-in AI are leading the pack, lowering costs and enabling the abilities that consumers are looking for. For instance, they might not know that their affordable robot vacuum is powered by on-device AI, with smart territory mapping abilities and the ability to avoid obstacles on the fly; they just know it works better.
The technology behind the curtain
While the technology is currently enabled by 4G, WiFi, and Bluetooth, what makes on-device AI so exciting for developers is the rise of 5G, which will enable massive device connectivity. 5G will offer much higher data rates, reducing latency and cost-per-bit reduction, plus higher system capacity, and a seamless interaction between the cloud and the device.
Companies are recognizing the potential of on-device AI to make their solutions better, and harnessing efficient AI processing is the first and most important step. It means finding ways to bring powerful neural networks onto devices, which sometimes involves compression, optimization, or quantizing, or a combination of all three. Trained neural network platforms can also increase the performance and power efficiency of on-device AI, and companies like Google, with TensorFlow, and Facebook, with PyTorch, are helping shape the industry.
To learn more about how on-device AI is unlocking tremendous potential for innovative companies, changing how consumers think about their devices, from smart phones to home electronics, and how to make the right technology investments, register now for this VB Live event!
Don’t miss out!
- How AI-powered smartphone apps are igniting the consumer imagination — and why
- How on-device AI is changing the future of connected devices
- How developers, OEMs, and ISVs can take advantage of on-device AI processing
- The technology you need to drive the best performance of on-device AI
Gary Brotman, Senior Director Product Management, Qualcomm Technologies
More speakers announced soon!
Sponsored by Qualcomm Technologies, Inc.