Why you should and how you can inform your chatbots with custom data or Wikipedia access

Extending my purely web-based, GPT3-powered chatbot to know content fed by me or that it retrieves automatically from Wikipedia

I showed you earlier how by informing your personalized chatbot with custom-provided information, it can better answer questions rather than just making up stuff:

Build ChatGPT-like Chatbots With Customized Knowledge for Your Websites, Using Simple Programming

You can use this approach to, for example, allow a chatbot on your personal or company’s website to “know” about you or your company and reply accordingly:

The web app chatbot answering to some questions about me by using information provided in “Luciano’s texts”, active on the bottom left.

Now, I show you another way to make bots smarter: by allowing them to search Wikipedia, extract information from it, and then reply using this information.

See how this bot first made up a wrong answer to the question “who is mastropiero?” when left to work only with its GPT-3-based “brain” plus information from “Luciano’s texts” (which doesn’t include anything about Mastropiero, so it’s useless for this question). And see the bot then answered my question correctly when it used Wikipedia as a resource (second answer):

Where to find this Wikipedia-enabled bot

This is essentially the same chatbot I described here, but extended to make use of the Wikipedia option (which wasn’t working in my first prototype):

Here’s my ChatGPT-like bot for you to use -and possibly install in your website customized to your…

Next? PubMed for more accurate replies on the natural sciences