Struggling to learn AI — any beginner-friendly courses?

Posted by FewCockroach2590@reddit | learnprogramming | View on Reddit | 14 comments

I’ve been trying to get into AI/ML, but I feel a bit stuck.

I started a course on HarvardX, but honestly it felt too advanced for me and I couldn’t follow everything properly.

I already know Python and can build projects, and I’ve understood some basics like search concepts, but I struggle with how to actually use the methods in practice.

When it comes to things like models, training, or applying what I learn, I get lost.

I’m looking for something more beginner-friendly that explains things step by step, not just theory.

Any courses or resources you’d recommend?

Also, should I focus more on theory first or just start building small AI projects?

[-]

DevilKnight03@reddit

Based on what people discuss in forums, the gap between “understanding” and “doing” is where most beginners struggle. Courses that focus only on theory don’t always show how models are actually built or used.

A common suggestion is to pair conceptual courses with something more applied. In those comparisons, udacity tends to come up for step-by-step projects, while platforms like coursera are mentioned more for structured theory.

[-]

Substantial-Peace588@reddit

Since you’ve already got a handle on Python, it might be worth shifting your focus to something a bit more hands-on and practical. That’s where things usually start to click, honestly. You could take a look at H2K Infosys their AI training is structured in a way that walks you through things step by step, and it’s pretty project-driven too, which helps a lot when you’re trying to actually use what you learn.

They also offer placement support, which can be a big plus if your end goal is landing a job. At the same time, don’t just rely on the course alone try building small projects on your own as you go. Even simple ones make a difference.

That mix guided learning plus getting your hands a bit messy with real work tends to speed things up in a way theory alone just doesn’t.

[-]

therealpolecat@reddit

What actually helps is switching to resources that walk you through the full process: load data, train model, evaluate, tweak. Once you see that loop a few times, things start to click. Udacity’s AI/ML tracks walk through that workflow step by step with projects, which can help if you feel lost doing it solo.

[-]

EfficientNoise215@reddit

If you’re struggling to learn AI, start with something simple and structured.

H2K Infosys is a good beginner-friendly option because:

Easy explanations (good for non-tech learners)
Covers basics like Python + Machine Learning
Includes hands-on practice and support

Best for building a strong foundation before moving to advanced AI topics.

[-]

Simplilearn@reddit

For a beginner-friendly and practical course, you can start with the Machine Learning Basics Free Course by Simplilearn. It covers core concepts like regression, classification, clustering, and model training with hands-on exercises, so you actually apply what you learn.

If you want to go further after that, their Machine Learning using Python course focuses on real workflows like data preprocessing, model building, and evaluation across multiple algorithms

Build small projects while learning. That balance is what moves you forward.

[-]

EducationLimp7068@reddit

you should start learning Neural networks, they are the core of how every modern deep learning and AI algortihm works, and you do not need any prior knowledge for it. just basic python and very fundamental knowledge of differentiation. below video is created by Andrej Karpathy, co-founder of OpenAI, which explains how neural networks work.

here is the video link: the spelled out intro of neural network and back propagation

this video is amazing but there is one problem when you actually try to follow along you end up pausing every few secs to understand what is being taught, then pause again to write code, then you lose where you were. its such a bad experience and because of this only like 2-3 percent of people actually finish the project.

i faced the same issue so i broke the whole video down into small lessons, each one focused on one concept and after understanding it you write code from scratch before moving on. built an AI tutor around it that draws and speaks like a real tutor at a whiteboard.

happy to share the link if anyone wants to try it.

[-]

Miserable-Whole592@reddit

Lo que estás describiendo es un punto en el que se queda mucha gente cuando empieza con IA/ML. No es tanto un problema de falta de conocimiento, sino una brecha entre entender los conceptos y saber aplicarlos. Cursos como el de HarvardX suelen ser más teóricos o con un enfoque académico, así que aunque entiendas partes, no siempre se traduce en construir algo práctico. Por eso da la sensación de “lo entiendo, pero no sé usarlo”.

En mi experiencia, hay dos formas de abordar esto:

1) Seguir profundizando en teoría, funciona, pero es más lento y a veces se siente desconectado de la práctica.

2) Pasar a aprendizaje aplicado. Empezar a construir cosas pequeñas, aunque sean simples: usar APIs, conectar modelos con inputs/outputs reales, crear pequeños workflows....

Esto suele cerrar la brecha mucho más rápido. La clave no es “aprender más”, sino aprender de una forma que te obligue a usar lo que ya sabes. Además, no todos los cursos están pensados para lo mismo, algunos son para entender ML en profundidad y otros están enfocados en aplicar IA en casos reales.

Y mucha gente no se da cuenta de que tiene que elegir ese enfoque. Yo pasé por una fase parecida y estuve investigando bastante diferentes formas de aprender, porque esa sensación de atasco es muy común. Si quieres, te puedo contar lo que me ayudó a salir de ahí.

[-]

LookTurbulent426@reddit

Look at the 3blue1brown ml videos, thats how I started. It’s not too deep nor does it tell u all u need to know but it makes understanding the other more advanced “harvardX” stuff easier. Gives you a little bit of intuition so you kinda know whats going on and have a basline of knowledge to fallback on.

[-]

dadesigns41@reddit

Me too. I think the problem is the tech is evolving faster than we can learn it. Bytebyteai course looks pretty dope but it’s 3K 🤣🤣🤣🤣.

I’m in maestro.org ai course….but it teaches from scratch so Im in like redundant python 101 as of now but ai courses are coming up. IK LLM RAG transformers MCP are all non-negotiable learning topics.

[-]

John_8PM_call@reddit

Learning Machine Learning (AI) isn’t really a basic beginner’s thing. It requires advanced university level math like Linear Algebra as a prerequisite. Between the math stuff and the coding stuff, it could take you 4+ years to really learn. That being said, if you’re still up for it, here are some resources:

https://reddit.com/r/learnmachinelearning/wiki/index

[-]

Glittering_Poem6246@reddit

Your problem seems not the language but project lifecycle understanding.

When you make a project especially a complex one you don't just start you have to think about what classes or methods would come in play.

But by far, making projects is better than a course - preferably a project you thought of.

[-]