Learning AI, One Spreadsheet at a Time: A Hands-On Approach
Artificial Intelligence (AI) often feels intimidating to those without a programming background. But what if you could learn its inner workings — softmax functions, backpropagation, even transformers — using a tool as familiar as Excel?
The “AI by Hand” GitHub repository by Prof Tom Yeh takes an innovative approach to teaching AI concepts by breaking them down into Excel-based exercises.
Here’s what it offers:
🔹 Foundational Concepts: Explore softmax, leaky ReLU, and other key functions.
🔹 Advanced Models: Dive into multi-layer perceptrons, RNNs, LSTMs, ResNets, and transformers.
🔹 Interactive Workbooks: Hands-on exercises for dot products, matrix multiplications, and linear layers — no programming required.
Why Excel?
By using Excel, the repository transforms abstract AI ideas into transparent, step-by-step workflows. Want to see how backpropagation calculates gradients or how attention mechanisms prioritize data? It’s all laid out visually, cell by cell. This demystifies the black-box nature of AI and brings a new level of clarity, especially for learners and educators.
The project is open-source under the MIT License, so whether you’re a student, teacher, or just curious, it’s ready for exploration.
Sometimes, learning AI doesn’t require code — just curiosity and a spreadsheet :)
Would this approach help you understand AI better? Let me know your thoughts!
