Introducing God's Translator

The excitement around Artificial Intelligence (AI) is contagious. Ever Curious Corporation, a startup I co-founded in 2015, was based on AI. A lot has changed since then. One of the most significant technical achievements has been the development of Transformer technology, which enabled the creation of Large Language Models (LLMs) like GPT (Generative Pre-trained Transformer).

I’d like to take a moment to offer a cursory explanation of how LLMs work. 

In simple terms, when you ask an LLM a question, it understands the question using the Transformer model, deep neural networks, and other methods for context, syntax, and semantics. It then responds based on this understanding.

Before LLMs can be useful to humans, they have to be trained. There are three main techniques for training and enhancing the performance of an LLM:

  1. Fine-tuning: This process involves further training a pre-trained LLM on a specific dataset to improve its performance in particular tasks or domains. Humans often review the LLM's outputs and adjust the parameters to enhance accuracy and relevance.

  1. RAG (Retrieval-Augmented Generation): This technique enhances responses by integrating an external database or knowledge base with the LLM. The system retrieves relevant information from the database and uses it to generate more accurate and contextually relevant answers.

  1. Prompt Engineering: This method involves designing and optimizing the prompts given to the LLM to elicit specific types of responses. By adding certain words or phrases to user queries, the LLM can be guided to respond in a desired way.

These three techniques can be used independently or in combination. 

I am curious (and own 🙂) and fortunate to have Laurent Parenteau and Heather Cooper as my friends. We decided to learn more about LLMs by building a product. We built something that would be fun to use with a possibility of increasing social harmony. 

Checkout - Ask God anything, anonymously. 

The following are a few frequently asked questions about God’s Translator : 

  1. Why should I use God's Translator instead of ChatGPT? 

    1. It is a more personal and enjoyable experience because: 

      1. All questions are anonymous and no data is stored. 

      2. You get the response in God’s voice. 

      3. We provide responses from all three Abrahamic religions: Christianity, Islam, and Judaism.

      4. We provide hyperlinks to religious texts so you can delve deeper into the answers.

  1. What LLM are you using? 

    1. Meta Llama 3 

  1. How does it work? 

    1. We are using a Python framework with Llama 3 on AWS. Simply, when you type your question on our page webpage ( which is using streamlit), we add a system prompt to the questions and generate a REST API call to our backend where Llama 3 gets the question and returns the response back to the webpage. 

In other words, we are using Prompt Engineering. 

  1. How do you plan to make money? 

    1. We don’t. We built the product to provide people with multiple perspectives on the deeply interesting questions. Our hope is that increases religious harmony. 

  1. Why include only three religions in the God’s Translator response? 

    1. Abrahamic religions are interconnected and believe in the same God. Other religions have different ideas of God.

Laurent leads engineering and Heather leads design. was created in less than one month of part-time work :) 

What would you like to ask God today? 

Popular posts from this blog

Obituary: Charles T. Munger

Systems Thinking as taught by Ackoff

Marry the best who will have you and other wisdom from Munger and Buffett