This application seeks to answer your question by having ChatGPT parse your Google search results for you.
It searches your question, goes to the first number of URLs from that result, scrapes the paragraphs from those webpages, and then asks ChatGPT your question with that web data as context information.
Using Google searches can help give ChatGPT more recent of info.
Here is an example of a question that is too new for ChatGPT:
And here it is being answered correctly, since GPTSearch has access to Google search results:
It can also help ChatGPT results to be more concise.
Here, ChatGPT does not infer the necessary context:
Whereas GPTSearch inherently understands context such as who is most relevant because of Google:
ChatGPT can also simply be wrong or silly, which Google search verification can help solve.
Here, ChatGPT gives a somewhat insane piece of Rust code, using very non-Rust style:
Whereas GPTSearch both gets info from real human Rust coders live from the web and is forced to be concise:
There should not be any special setup other than standard Cargo Rust tools.
The runtime parameters are your search query and an appropriate OpenAI API key.
So run with a command similar to the following:
cargo run "What temperature should I cook chicken until?" Sk-abcdefghijklmnopqrstuvwxyz
Feel free to contact me at [email protected] with any questions / inquiries.