Totally agree! For a budget pick, Id say check out Sololearn: Learn to Code cuz their AI path is basically free and perfect for quick commute sessions. Gl!

Stumbled upon this! Quick question - do u eventually wanna write code or just get the 'why' behind the math? Honestly, I've been super happy with these:

• Brilliant.org - Uses interactive puzzles to explain backpropagation and weights visually.
• Enki: Learn to Code & AI - Perfect for bite-sized logic during commutes.

Basically, Brilliant focuses on visual intuition for gradient descent, which is highkey the hardest part. Worth the sub imo!

Just caught this thread today. > Ideally, I’d love an app that breaks down the underlying math into bite-sized pieces and maybe even uses gamified quizzes to help the information stick. Tbh, I’ve been a DIY learner in this space for a long time, and I’ve found that the "fun" apps sometimes over-simplify the math to the point where its hard to actually apply it later. Before you sink money into a subscription, I gotta ask—how much math are you actually comfortable with right now? Like, are we talking "I remember high school calculus" or "I haven't seen an equation in a decade"? It makes a huge difference in which path you should take. If you really want to learn the *logic* without just playing a game, I’d say just stick with any of the big university-affiliated platforms. You basically can't go wrong with those academic brands because they prioritize the actual theory over flashy UI. It’s a bit more of a grind, but your better off in the long run if you want to actually understand how these weights and biases are shifting under the hood.

TL;DR: Don't just focus on the logic, make sure the app teaches you how to benchmark and test performance properly or you will get stuck in the beginner trap. Honestly everyone is giving great advice on the visual side of things but as someone who obsesses over real-world performance I have to give you a bit of a warning. A huge mistake beginners make is getting a high score in a gamified app and thinking they have mastered ML. In the real world performance isnt about just getting the logic right once it is about rigorous testing and understanding benchmarks. If you want that click moment you need to find resources that explain why a model that looks perfect on paper might actually fail when it hits real data. Most mobile tools use curated datasets that are way too clean. You might feel like a genius when your model hits 99 percent accuracy in a quiz but that is usually just overfitting and it wont work on a different dataset.

• Coursera mobile app (the testing and validation modules are top tier)
• Kaggle (use the mobile browser to read through winning competition notebooks to see how they benchmark)
• Elements of AI (solid for a high-level performance overview) Basically just watch out for apps that skip the how do we know this actually works part. If they dont mention cross-validation or F1 scores you are only getting half the story.

oh man, I feel u so much on that commute struggle!! Honestly, I was in your shoes a few years back—totally wiped after work but dying to understand AI. I tried reading documentation on the train and it was a total disaster lol.

For your situation, I highkey recommend these:
* Brilliant.org iOS App - This is literally THE one. It breaks down neural networks and linear regression using interactive visuals that actually make sense. I’ve used it for years to stay sharp.
* Enki: Learn Code & Data Science - This one is also great for bite-sized daily goals and has a solid ML track if you want something more structured.

My biggest lesson after years of messing with this stuff? Don’t worry about the complex syntax yet. Basically, you gotta focus on the 'why' behind the math first. Once the 'big picture' clicks in ur head via these apps, the actual coding part becomes way less intimidating. Brilliant really is worth the sub if you want that gamified feel. gl with the learning!! 👍

So I totally agree that the interactive visual approach is basically the gold standard right now for building intuition without getting bogged down in syntax. Looking at the broader market from a pedagogical standpoint, there's a massive divide between apps that prioritize "engagement" and those that maintain technical rigor. As someone who works in the field, I’m always a bit cautious because many consumer-facing platforms oversimplify things like backpropagation or the stochastic nature of these models to the point where they can be misleading. Before you sink money into a sub, I'm curious about your end goal—are you looking for high-level "AI literacy" to keep up with industry trends, or are you aiming for a foundation that eventually scales into professional model development? Also, what's your actual comfort level with the underlying linear algebra? Tbh, most apps on the market right now tend to gloss over the math, which can be risky if you want a truly deep understanding of how these architectures function.

Late to the party but I wanted to chime in because I wasted a ton of time early on with apps that felt like games but didnt actually teach me how to measure if a model is actually good or just lucky. I remember being so frustrated when my first real project failed because I didnt understand how to tune performance or why my accuracy was a total lie. I would suggest looking at Coursera iOS App specifically for the Andrew Ng courses. Be careful tho, its way more rigorous than the typical gamified app, but thats exactly why it works. You might want to consider downloading the lectures for your commute so you can focus on the performance metrics sections... thats where you learn about things like overfitting and bias-variance tradeoffs which are the real keys to making ML work. I found that the big picture finally clicked when I stopped looking for "correct" syntax and started focusing on how to evaluate the logic. Its a bit of a steeper climb but honestly so worth it if you want to actually use this stuff later. Let me know if you run into any math hurdles!

I totally agree that building that visual intuition is the absolute biggest hurdle when ur starting out—if u don't 'see' how things like gradient descent actually work, the formulas just look like total Greek lol. Having hopped between different learning tools for a few years now, I’ve found that the 'big picture' really starts to stick once u bridge the gap between just 'playing a game' and actually writing out the logic. I spent a lot of time with Mimo: Learn Coding on my own commute, and it was sooo helpful for keeping those core concepts fresh. They have a dedicated AI track that’s super approachable for beginners. What I liked about the long-term experience with Mimo is that it forces u to engage with the Python syntax in a way that feels natural for a small screen, rather than just clicking through simple multiple-choice quizzes. It really helps the logic sink in because ur basically 'building' tiny parts of the model yourself. One thing to keep in mind though—from my experience, don't get too stressed if the heavy math feels a bit fuzzy at first. These apps are great for the vocab and the general flow, but it usually takes seeing it three or four different ways before the 'why' truly clicks. Just keep that streak going!