Artificial intelligence (AI) has already transformed many elements of our smartphone experience, from smart voice assistants that make it quick and easy to carry out tasks on your device to AI-enabled photography features such as AI noise reduction and smart High Dynamic Range (HDR).
But we are still in the early stages of AI’s development. Smartphone makers, software platform providers like Google, AI software firms like Meta and OpenAI, and fabless semiconductor companies such as MediaTek are all collaborating on the next set of powerful innovations.
The exciting news is that these advances may start out in the most expensive flagship devices but will soon be accessible to nearly anyone. IDC estimates that AI-enabled mobile phones accounted for 20% of total sales in 2024. In four years, that share is expected to rise to 70%.
Edge AI and NPU advances
There are two interlinked developments driving this growth. The first of these is the advent of large language models (LLMs) as well as generative AI (GenAI applications like ChatGPT, Google Gemini and Midjourney. GenAI is a new type of machine-learning algorithm that can generate original content based on data like images, videos, audio and words.
The second is the rapid pace of advancement in edge AI hardware, specifically AI-enabled systems-on-chips (SoCs). These SoCs enable smartphone manufacturers to create handsets that allow Gen AI applications to be processed inside your smartphone or on the cloud, depending on your needs and the app.
Handling some of your GenAI applications directly on your device has a range of benefits. Your applications can be faster and more responsive because you don’t need to send data backwards and forwards over the internet. You can also keep sensitive data, like voice or facial biometrics, on your device. Finally, your apps will work even if there is no network coverage.
The new generation of AI applications
This technology is ushering in a new age of innovation, creativity and productivity by democratizing GenAI. Many smartphone manufacturers are already using our MediaTek Dimensity platforms to create some powerful Gen AI applications for the edge and the cloud.
Here are a few examples of the new and upcoming GenAI applications to look out for:
- Google’s Circle to Search is a popular cloud Gen AI feature already appearing in many Android devices. You can instantly search for anything you see on your phone without needing to switch apps.
- Some smartphones already offer AI interpreters to overcome language barriers with real-time translations as well as AI notetaking to transcribe and summarize calls and meetings.
- AI Subtitles supports the recognition and translation of multiple languages in real time.
- Gen-AI Super Zoom creates high-quality zoom photos by using generative AI, while on-device video generation creates realistic animations directly on the phone, for example, transforming portrait photos into moving images.
- On-device GenAI models will evolve from static voice assistants into context-aware companions that understand tone, emotion and habits. They’ll anticipate needs, such as suggesting a reply, drafting a message or planning a route, based on your real-world context.
- Edge AI will power smarter, adaptive security that continuously learns from device behavior. It could detect unusual app activity, block phishing attempts, or even alert users to cloned apps or spoofed calls, all without sending sensitive data to the cloud.
On-device GenAI, powered by advanced MediaTek platforms, is at the heart of this transformation, bringing intelligence, efficiency, and personalization to billions of users. Our role is to push NPU technology forward and work with our partners to make GenAI accessible and available to all.
By Christo Wessels, MediaTek’s Manager for Sub-Saharan Africa
