Qualcomm
At Mobile World Congress 2024, Qualcomm is enhancing its portfolio of AI-on-phone capabilities facilitated by the Snapdragon series silicon for Android phones. The chipmaker has already showcased remarkable AI features for the Snapdragon 8 Gen 3 flagship, such as voice-activated media editing, on-device image generation using Stable Diffusion, and a smarter virtual assistant built on large language models. Now, Qualcomm is adding more power to these AI superpowers. One of the new additions is the ability to run a Large Language and Vision Assistant (LLaVa) on a smartphone. Think of it as a chatbot with Google Lens-like abilities. This means the AI assistant can not only process text input but also images. For example, you can show an image of a charcuterie board and ask questions about it. The LLM (Large Multimodal Model) with over 7 billion parameters can tell you all the kinds of fruits, cheeses, meats, and nuts on the board. It can also handle follow-up queries, allowing for a seamless conversation. While tools like ChatGPT have gained multi-modal capabilities, the difference is that Qualcomm’s solution processes everything on the phone, making it faster and with less privacy risk as the data is handled locally. “This LMM runs at a responsive token rate on device, resulting in enhanced privacy, reliability, personalization, and cost savings,” says Qualcomm. It’s not yet clear if the LLaVa-based virtual assistant will come as a standalone app or with a fee.
The next announcement from Qualcomm focuses on the creative side of image generation and manipulation. Not long ago, Qualcomm demonstrated the fastest text-to-image generation on a phone using Stable Diffusion tech. Today, they give a first look at LoRA-driven image generation. LoRA, short for Low-Rank Adaptation, is a technique developed by Microsoft. Training an AI model can be expensive, time-consuming, and resource-intensive. LoRA reduces the model weight by focusing on specific parts and reducing the number of parameters for training. This results in lower memory requirements, faster processing, and easier adaptation. The LoRA distillation technique has been applied to the Stable Diffusion model for efficient image generation. Qualcomm sees it as a perfect fit for smartphones, and even rival MediaTek has adopted the same solution for its flagship Dimensity 9300 chip. Qualcomm is also showcasing other AI tricks at MWC 2024, such as the ability to expand the canvas of an image using generative AI fill and AI-powered video generation. It will be interesting to see how they bring these features to smartphones and what kind of impact they have.

Qualcomm
Another important aspect of Qualcomm’s AI offerings is its focus on privacy and security. With the increasing concern over data privacy, Qualcomm ensures that all the AI processing on the phone is done in a secure and private manner. The use of on-device processing minimizes the risk of data leakage and gives users more control over their personal information. In addition to these advancements, Qualcomm is also working on improving the performance and efficiency of AI applications on smartphones. They are optimizing algorithms and leveraging the power of the Snapdragon chips to deliver seamless AI experiences. This includes faster inference speeds, better energy efficiency, and more accurate predictions. The company is collaborating with developers and software partners to ensure that AI applications can take full advantage of the capabilities of the Snapdragon platform.
Looking ahead, Qualcomm’s AI strategies are likely to play a crucial role in shaping the future of smartphones. With the continued evolution of AI technology, we can expect to see more innovative features and capabilities integrated into our mobile devices. Qualcomm’s efforts in this area are helping to drive the development of smarter and more personalized smartphones that can meet the evolving needs of users. Whether it’s for enhanced photography, intelligent assistants, or seamless multitasking, AI is set to become an integral part of our mobile experiences. As Mobile World Congress 2024 progresses, we’ll be eagerly watching to see what other exciting announcements Qualcomm has in store for us in the world of AI and smartphones.

Qualcomm
In conclusion, Qualcomm’s advancements in AI at Mobile World Congress 2024 are significant. The company’s ability to combine powerful AI capabilities with on-device processing, privacy focus, and performance optimization sets it apart in the smartphone industry. The introduction of the LLaVa and LoRA-driven image generation showcases Qualcomm’s innovation and commitment to pushing the boundaries of what is possible with smartphones. As we move forward, it will be interesting to see how these technologies are further integrated into our daily lives and how Qualcomm continues to lead the way in the development of AI for smartphones. Whether it’s through improved user experiences or new applications, Qualcomm’s AI efforts are likely to have a lasting impact on the way we use our mobile devices. With Qualcomm at the forefront, the future of smartphones and AI looks promising.