Exploring OpenAI
Originally posted — https://setkyar.com/exploring-openai
As many of us have probably seen, chatGPT has become very popular recently, with people sharing conversations on social media. I was also interested in chatGPT, so I tried it out and was impressed by its capabilities. Since then, chatGPT has become like a technical friend to me. Whenever I have questions, especially about technical topics, I turn to chatGPT instead of Google. It’s been very helpful in my day-to-day work, especially when I need to remember Linux commands or learn about new technologies.
I became interested in exploring OpenAI further and started asking chatGPT a lot of questions. I discovered that there is an API available and had the idea to create a Python program that could call the API directly, rather than using the chatGPT UI. I asked chatGPT to create such a program and then modified it slightly. I set up the program with an alias called “ask” so that I could easily use it by typing “ask” followed by my question. It has been very useful and fun to use on a daily basis.
As I continued exploring, I wondered if I could create a chatGPT-like system that worked like Google Assistant, with voice commands and OpenAI responses through voice output. I asked chatGPT to create a simple web program that could take voice input, convert it to text, pass it to the API server, and then call the OpenAI API. When the API responded, the web frontend would convert the text to voice and read it out loud for me. While it was fun to create this system, it wasn’t very practical to use. However, I believe that if I could use other OpenAI APIs, such as whisper, it could be a great tool. I’m excited to see what other AI-based applications and services will be developed in the future to improve our productivity. You can find the code for the “ask” alias, the voice assistant, and other projects I’m working on in my GitHub repository.
FYI — This blog article has been optimized by chatGPT.
Edited. I created a CLI base application. You can check it out on my GitHub. Easy to install as it’s binary.
