From Rule-Based AI to Generative AI: The Emergence of Personalized AI Tools for Everyone

The year 2023 marks a significant breakthrough for AI, particularly with the emergence of large language models like GPT from OpenAI. This shift marks a transition from "reactive AI" to "generative AI." According to Market Data Forecast, the global AI market is set to expand from $150.2 billion in 2023 to $1.3452 trillion by 2030, boasting a remarkable compound annual growth rate of 36.8%.

In this context, various domestic large models in China—such as Baidu’s Wenxin Yiyan, iFlytek’s Spark, and Alibaba’s Tongyi Qianwen—are revolutionizing our work and daily lives. However, the development and training of these models rely heavily on vast amounts of data. Simply put, the volume of data determines the size of the model's parameters, which in turn influences its capabilities. Processing this data also requires substantial computing power, necessitating advanced hardware.

As a result, the most capable models, which can generate images, videos, and even audio tailored to specific requests, primarily operate on cloud servers. The training data used for these models is typically sourced from users who have signed usage agreements, ensuring the legality and standardization of the data. Unfortunately, this often results in responses that are generic and lack personalization, failing to cater to individual user needs or unique scenarios.

Given this challenge, models with relatively fewer parameters that can handle simpler, more personalized, and private tasks are becoming an increasingly attractive direction for AI development. These models can be deployed locally, allowing users to train them with their own data directly on their devices.

Network Attached Storage (NAS) has gained attention as a promising AI hardware solution that can accommodate such models in everyday consumer life. Serving as a personal data hub, NAS not only provides storage but also delivers essential computing and network capabilities, all while enhancing data security. This makes it an ideal platform for localized AI applications.

However, even in an era of abundance, owning a personal NAS can seem out of reach for many. The relatively high cost of hardware, combined with the complex set-up process that can overwhelm those without technical expertise, creates a steep learning curve. Consequently, traditional NAS systems may pose an entry barrier for average users.