A few days ago, I had the honor of being warmly invited by 360 to attend the ISC.AI summit they organized. In the breakout room of the summit, I had an in-depth exchange with Orange, Kazhik, Kun Dao, and Lin Yi's Xiaobai about AI products.
In fact, although everyone is related to the field of AI, there are obvious differences in the sub-areas we focus on. Take Orange, for example, he is more concerned with the application of efficiency tools; while Kun Dao, being a video creator himself, naturally focuses more on AI applications related to video creation. It is precisely because of this difference that we were able to collide many interesting and enlightening sparks of thought during the exchange.
However, through the exchange, we reached a consensus that the penetration of AI in various fields is still not ideal. We believe that to further promote the widespread application of AI, on the one hand, we need to attract more high-level creators to join the application and development of AI, and on the other hand, we need to continue to work hard to reduce the usage threshold of AI products, allowing more ordinary users to easily get started.
On the day of the summit, 360 released an eye-catching AI assistant. This AI assistant integrates all the LLM models provided by 15 model manufacturers nationwide. During the development process, 360 first conducted a detailed and rigorous test of all models to clarify the capabilities and characteristics of each model. Then, they made full use of their own search intent judgment capabilities and advanced AI model routing technology, so that when users ask questions, they can accurately call the model with the best performance in the corresponding field to answer. Surprisingly, after actual testing and verification, this AI assistant, which combines the advantages of many models, has achieved results that surpass GPT-4o comprehensively.
That morning, I shared information about this product in my circle of friends. To my surprise, those who showed a strong interest in this product were not my colleagues in the industry, but those ordinary friends who only had a general understanding of AI before. Many of them even came to me to ask for the product's address, and their enthusiasm far exceeded my expectations.
It was not until this moment that I truly realized how strong the demand for an excellent and comprehensive AI model is among ordinary users. In the current market situation, due to various problems with domestic models, there has not yet been an AI product that is comprehensive enough in terms of capabilities. However, the needs of users are real, and not everyone can use the best-performing models as smoothly as we industry insiders can.
Finally, we once again emphasize the core issue of user needs. When we develop and promote AI products, we should first consider how to solve practical problems for users and put their needs first, and then think about the selection and application of models. We should make full use of the models we have, but at the same time, we cannot be limited to our own models. We should integrate more resources with an open attitude to provide better services for users.