I'm 16 and have recently become interested in programming and artificial intelligence, but I've never programmed before. I originally wanted to build my own voice assistant—something like a personal "Jarvis"—as a long-term project to learn programming, APIs, software architecture, and technical design.
I've used an AI tool to help draft requirements and an initial architecture, but I now want to do more of the research and decision-making myself. Before I go any further, I'd like advice on which programming language to start with, what books, courses, or other resources would be useful, and what questions I should be asking when planning a serious software project.
I'm also considering using the project as part of my college applications. I don't want it to look like I simply connected a few existing AI APIs or copied generated code. How can I make the project genuinely educational and distinctive while demonstrating that I understand the software I'm building? What smaller projects or learning path would be sensible for someone starting from zero?
3 Answers
Start by learning programming itself rather than designing the complete assistant immediately. Python is a sensible first language because it has approachable learning materials and a large ecosystem for later AI work. An introductory course with exercises—such as a university-style Python course—would be more useful than having an AI generate a roadmap or code for you.
Build small programs first: a command-line calculator, a text adventure, a to-do list, a file organizer, a simple game, or a program that uses a public API. Each project should be written and explained by you. Gradually add testing, error handling, documentation, and version control. After that, you’ll be in a much better position to decide what an assistant should actually do and how its components should fit together.
You’ll eventually need more than Python: basic data structures and algorithms, debugging, operating-system concepts, networking, databases, software design, and possibly linear algebra and probability if you want to study machine learning in depth. You do not need to master all of those before writing your first useful program, though.
Learn them in stages. Focus first on writing code without relying on generated solutions, reading documentation, and figuring out errors through your own research. Later, study how neural networks and language models work mathematically. The important distinction is between building an application around an AI model and implementing or understanding the underlying model itself.
So the realistic approach is to keep the assistant as a long-term direction, but use smaller programs to learn the skills needed for it. I was definitely trying to jump ahead by calling the planning stage a technical design stage before I understood the basics.
A personal assistant can be a useful long-term goal, but it won’t teach you how an AI model works if you only connect services built by other people. Treat it as a sequence of projects: first a normal command-line application, then perhaps a program that stores and searches notes, then one that calls an API, and only later a voice interface or language-model feature.
Keep a development journal and explain your design choices, limitations, failures, and changes. If you eventually use an existing model, be honest about what you built yourself and what came from an external service. A thoughtful, well-documented smaller project is more convincing than a flashy assistant assembled from generated code.
That makes sense. I’ve started an introductory Python course, but I also want to practice by making small programs instead of only following lessons. I’ll put the assistant idea on hold while I build those fundamentals.