Book Detective · Joelbooks’ AI-Powered Book Finder

5 min readDec 6, 2023

What Is Book Detective · AI Book Search?

Book Detective utilizes the power of artificial intelligence to search for books. It efficiently works even if you only remember fragmented details or if there are multiple book elements that you are looking for. Simply tell the tool what kind of book you are looking for.

Visit Book Detective

Let’s consider a sample search: “I’m in the mood for a fantasy novel featuring dragons and epic romance with a strong female protagonist who is very clever.”

What databases are you using?

We utilize two primary data sources. First, we use OpenAI’s LLM, which is arguably one of the largest AI models currently available. Second, we utilize our website’s dataset, where authors have directly submitted excerpts and descriptions of their books or our editors wrote about the book in a form of a list article. We also opened the opportunity to include books directly to our AI dataset.

How good is our tool?

We did our first experiments with the alpha version of our tool called “What is that Book” several months ago.

We compared our tool with BookAbout’s and Findyournextbook’s search engine, which are described as using AI technology. The results speak for themselves.

After we moved our tool to ChatGPT interface in a form of a custom GPT called “Book Detective” we further improved the tool with the capability to find also freshly released or even upcoming book titles (these information isn’t part of ChatGPT’s knowledge base).

Here are some sample conversations with the beta version of Book Detective (Dec. 2023):

We did of course the Flowers for Algernon test with the new version too, check Book Detective’s answer:

Can I submit my book to the Book Detective database to make it searchable?

You can directly submit fresh new or older books to our AI dataset.

Are there any limitations when using the beta search?

Book Detective is currently available for ChatGPT Plus members. With the release of GPT Store it will most likely available free for the public (at least we won’t ask for any fees using it).

AI Technology Behind Book Detective

The current version of our tool is GPT-4 based and added several extra book specific dataset curated by our team. These custom dataset include more than 4,500 fresh book titles which will be extended hopefully with further 100,000 titles in the future. It’s important to note that our goal is to include only quality works in this database, ensuring that our recommendations are both relevant and of high literary merit.

Large Language Models (LLMs) like GPT-4 are at the forefront of this technology. These models are trained on vast amounts of text data, enabling them to understand and generate human-like text. This capability makes them particularly well-suited for applications in book recommendations, where understanding nuances in literary themes, styles, and genres is crucial. The AI’s ability to process and analyze a large volume of books, reviews, and metadata allows it to identify patterns and correlations that might be missed by human curators.

We are committed to addressing the challenges that come with using AI in book recommendations. This includes mitigating any biases in the dataset and ensuring a diverse and inclusive range of books are represented. We understand the importance of exposing readers to a wide array of voices and perspectives. We continuously works to enhance the tool in this regard.

AI Book Search Engine’s Impact on Readers and the Publishing Industry

AI book searches aren’t new and there are a lot of engines out there, but we think that choosing the right platform is essential. ChatGPT proved that users like the chat based user interface, where you can freely explain what you are really looking for. For this purpose we developed a kind personality for Book Detective (Emily), which I think is essential to visit her regularly.

What is our standpoint on generative AI?

We are aware of the recent events in the book world and the backlash from using AI technologies. A lot of people exploit these technologies in a wrong way. On one hand, we also see a bunch of low-quality, fully AI-generated self-published books on Amazon where the writer hasn’t even bothered to read them before publishing. This is a completely unethical use case for AI.

We know that major AI companies use book datasets as inputs for their tools. The downside of generative AI is that it can effectively mimic someone’s writing style, which poses a problem.

However, on the other hand, this technology has the potential to do beneficial things, such as

  • performing highly relevant contextual searches, as is the case with our tool,
  • it can help improving the quality of the sentences we draft as authors,
  • it is capable to do quick edits, keeping the meaning of the content,
  • it is capable to illustrate non-illustrated books with image generation models,
  • the technology is capable to effectively summarize books,
  • LLM models will enable people to interact with books in a non-linear way in the future.

So, generative AI embodies both good and evil, and we should approach this technology with caution. I’m generally very positive about the future. I think there are still a lot of hidden benefits of AI, but we need to mitigate the downsides and filter out the AI junk. If you feel to make a comment, feel free to submit your feedback in the contact form.

Originally published at on December 6, 2023.