Learning AI-where to start

Not specially right now. Just don’t want to be the old guy that got left behind with technology. Kinda want a good base of it and who knows what I might learn and enjoy doing later on.

Will look into that.

I don’t know that I will use it in my job, but I would like to have some knowledge of how it works if it’s ever brought up.

quote:

---

start with a specific problem you want to solve and then use AI to solve that problem

---

This is how I've tried to learn it recently.

I run a factory. We have a software that powers our factory, and scanners within the factory tell me what order is where. I run a report several times a day to track production. I've never liked the report. It's a data dump and formatting is hamstrung by the software that gives it to me. I have to do a lot of sorting and manual calculations to get the info I want out of it.

So, I set Claude on it. It took me half a day of going back and forth with Claude to get things right, but man, now that it's working, what used to take me 15-20 minutes now takes me five seconds, and I was doing that multiple times a day.

In addition, Claude learned about my factory and processes along the way. Without prompting, it started doing additional analysis on the reports and comparing them every time I run it to raise some good questions.

All that to say, I agree with you. Find a problem unique to you, and jump in. I simply talked (typed, actually) to it like it was a person, and we went back and forth to get it right.

Python is mandatory to master. Focus on libraries like NumPy and Pandas for data handling. Don't build a model from scratch yet, as your learning. Learn how to plug OpenAI, Anthropic, or open-source Hugging Face models into a basic Python script using API keys.

Moreover, real-world AI models don't just read clean CSV files. They pull data from massive enterprise databases. You use SQL to clean, filter, and prepare millions of rows of data before feeding it into a machine learning model or vector database. Mastering SQL is a must for machine learning.

Learn RAG (Retrieval-Augmented Generation). This is how you give an AI model access to external data, for example, a company's internal documents. You’ll want to learn about Vector Databases like Pinecone or Chroma. Pair this with ETL in cloud computing you’re golden. Data lives everywhere in Excel sheets, user activity logs, CRM software, SQL databases, and third-party APIs. In the Extraction phase, you read and pull raw data from all these different places without disrupting the main business operations.

quote:

---

Python is mandatory to master

---

I've found that certain AI tasks work much better when I request Python code to execute a task rather than let AI try the task. To my surprise AI agents can be really bad at counting and iterating. Request AI to create a picture of 5 cats. Then ask for modifications like remove 2 or add three. The results can be all over the place.

I've found Gemini quite good at tutoring me. Just ask how to improve prompts, format prompts, limit hallucinations, etc.