Creating a Non-AI Database That Understands User Input

If you're looking to build something that functions like an LLM but without actually using one, you might need to adjust your expectations a bit. It's tough to create a system that doesn't at least mimic some aspects of AI-based learning while avoiding all of its potential pitfalls. Maybe simplify the scope and aim for a prototype that gets the basics down first? Keep in mind, LLMs are not sentient and just work off the data they were trained on.

You might want to explore the idea of creating a knowledge graph that links entities and concepts. This involves a lot of groundwork, like establishing relationships between different topics (e.g., linking 'honey badger' to 'animal' and going from there). It's definitely a complex project, and while it's possible to build something functional, capturing the full range of user queries accurately without AI will be challenging.

What you really want might not be a traditional database. It sounds like you're aiming for a conversational system that avoids the issues seen in many AI chatbots, like incorrect responses. You could consider using 'intents' where you provide a variety of sample phrases and develop a logic for how those phrases connect to actions. This method is similar to how older smart assistants worked, but even that approach has limitations without machine learning.

Just a heads up—if you're looking to parse natural language, it gets messy quickly. Realistically, you can build a system that performs well in limited areas (like specific subjects for homework), but trying to cover every possible topic in a meaningful way without machine learning will feel clunky. User inputs can vary significantly in phrasing, and the complexity of human language might create challenges you're not anticipating without AI support.

To tackle this, you could consider tokenizing the user input in a mathematical way that allows your program to recognize patterns or commands. The challenge lies in defining what 'understanding' means for your application. What action do you want the system to perform based on inputs like the honey badger example? It might be helpful to clarify the purpose of your app further so that you can get more targeted feedback.