Review of the Week#
Originally, I thought it would be completely achievable to write a weekly report once a week, but life has its completely unpredictable characteristics. Those past agreements, in this endless change, are destined to be unpredictable.
After the release of the gpt-3.5-turbo API, various applications based on it emerged like mushrooms after rain. Seven out of ten tweets were about new applications. This situation made me start thinking, besides being a user who benefits from this wave of artificial intelligence, can I do something creative and use this powerful platform to create my own products?
Soon, a brilliant idea came to my mind: a browser extension that provides the ability to access web pages and user input, and then uses the user's custom prompt and key to call the API to process the provided information, with the user accepting the results returned by the API. This extension acts as an intermediary layer between the browser, the user, and the API, and meets the personalized text processing needs while browsing the web by writing, presetting, and sharing prompts. The reason why all this can be achieved is the extremely low threshold of natural language. I define this extension as the Playground for ChatGPT platform, which is a platform for learning "programming," except that the programming language is natural language, and the entire process from development to product is completed during the learning process. All the user needs to do is write the corresponding prompt based on their needs and define the content for output and input.
Of course, during the recent testing of the API, I found that this model has some limitations. It is not stable for processing results that require precise output, which means that many functions cannot be implemented, and although the cost has been reduced, it is still not low enough. However, I believe that these problems will be solved through continuous iterations.
So, the reason why I am writing this weekly report now is because of this. It is difficult to extinguish the enthusiasm once it starts to burn.
However, now I have calmed down and temporarily put it aside. This is because I have realized that I cannot complete it in the short term. The complexity of its implementation has exceeded the limits of what I can handle at the moment. I can't even use the most basic technology stack to complete it. With only a basic understanding of JavaScript, I have to give up the idea of quickly creating an MVP because what I can produce with my current skills may be completely unusable and will incur higher costs in later maintenance. There is still a lot for me to learn.
I will get back on track tomorrow, and I am really looking forward to this time.