The AI Chatbot that can Cite its Sources

The AI Chatbot that can Cite its Sources

The Atlantic just wrote an article about how generative AI can't cite its sources. This might be true for OpenAI, but it is not true for an AINIRO AI chatbot.

Citations is actually one of our core features, and we've got some partners who became partners because of our ability to deliver citations. In some industries providing citations is a must, such as for legal AI chatbots for instance.

One of our partners is using our chatbot technology to deliver a natural language chatbot based upon hundreds of PDF files. These files contains information about regulations and employment law, and it is therefore imperativ that their chatbot can deliver citations, to allow their users to verify the information, and also look up which regulations and laws are relevant for their questions.

Other partners are using our technology more like AI search, and instead of focusing on its ability to deliver natural language answers, came to us because we can use natural language queries to funnel users into specific pages in their website, allowing their users to rapidly find information. Watch the following screenshot of how I get citations from our AI chatbot by simply asking it to provide citations.

Our AI chatbot delivering citations

Citations - A Crucial Feature

About 50% of our partners would never have been able to use our technology without this simple feature. Citations are crucial in areas such as.

  • The scientific field
  • Legal AI
  • Medical AI
  • Journalism
  • And yes, even the E-Commerce field

The first 4 above might be easily understood, but even an AI chatbot that functions as a "simple sales assistant" needs citations to funnel its users into purchasing pages. So at least when it comes to our technology, The Atlantic is 100% wrong. You can try it out below.

Wrapping up

The Atlantic might be right in that vanilla ChatGPT cannot deliver citations. This is because ChatGPT is built exclusively upon generative AI. Our technology adds RAG to the mix. RAG implies Retrieval-Augmented Generation. Basically, with RAG we can "seed" the underlying AI with additional information, helping it answer its questions. This eliminates a whole range of problems such as for instance.

  • We can deliver AI chatbots that don't hallucinates
  • We can deliver AI chatbots that can answer questions ChatGPT don't know how to answer
  • We can deliver AI chatbots that are experts in one particular field, and never answers questions not related to its field
  • And yes, we can deliver AI chatbots that are able to deliver citations

If you are interested in hearing how we can help you with AI, with or without citations, we would love to hear from you.

Thomas Hansen

Thomas Hansen I am the CEO and Founder of AINIRO.IO, Ltd. I am a software developer with more than 25 years of experience. I write about Machine Learning, AI, and how to help organizations adopt said technologies. You can follow me on LinkedIn if you want to read more of what I write.

Published 30. Jun 2024

How I built a Billion Dollar Company - Alone!

Neutral evaluations at this point in time, would probably evaluate AINIRO to one billion dollars, implying mission accomplished.

Read More

Long Term Memory AI Chatbot

Building a long term memory AI chatbot with AINIRO is easy. Everything can be accomplished using No-Code constructs.

Read More

The End of Google

Google will soon die. They've failed over and over again with AI, and before they'll have anything basic working, every single independent software company on Earth will have integrated something else.

Read More