Smartphone manufacturers are competing to lay out AI large models, is it a technological breakthrough or a marketing gimmick?

In the current commercial haze of the technology industry, AI large models have become a new opportunity that mobile phone manufacturers are eagerly pursuing. However, is this seemingly glittering opportunity really a breakthrough for the industry, or is it merely a marketing gimmick?

Recently, several domestic smartphone giants have announced their foray into the field of AI large models. Xiaomi showcased its self-developed 1.3 billion parameter large model at its annual launch event, claiming it could rival the 6 billion parameter cloud model in certain scenarios. Huawei plans to deeply integrate the "Pangu Large Model" into the HarmonyOS 4 system. OPPO launched the "Xiao Bu Assistant" based on AndesGPT technology, while vivo will unveil its AI large model and new operating system at the upcoming developer conference.

These smartphone manufacturers are all emphasizing the excellent performance of their large models in authoritative evaluations, seemingly preparing for the upcoming "AI smartphone" battle. However, running large models on mobile devices faces numerous challenges, including memory usage, computing power, and power consumption issues. To address these challenges, most manufacturers have adopted a cloud-edge collaborative strategy.

MediaTek has partnered with OPPO and vivo to jointly develop a lightweight deployment solution for large models on the edge. This edge-cloud integration approach can meet user needs in various aspects such as computing power, performance, energy consumption, and privacy protection, and is considered one of the preferred strategies for integrating mobile phones and AI large models.

However, there are also doubts within the industry regarding the use of large models running on mobile devices. Some believe that this is more to cater to short-term market trends rather than a genuine technological breakthrough. To adapt to mobile hardware, large models have to undergo significant compression, which may lead to a substantial decline in model performance.

Moreover, it seems that mobile phone manufacturers are currently too focused on applying AI large models to voice assistants, and whether they have truly considered the actual needs of users is open to discussion. The true popularization of AI large models in the mobile phone sector still faces many challenges, and the current efforts may just be the beginning of exploration.

In the future, smartphone manufacturers need to find a balance between chasing technological trends and meeting actual user needs. It will take time to verify what kind of innovations AI large models can bring to smartphones. Regardless, this new transformation in the smartphone industry initiated by AI large models will undoubtedly have a profound impact on the entire industrial chain.