Ask HN: How should I spend 10 weeks delving into AI?

[thelock85]

For context, I have done a lot of hacking mostly in JS throughout the years, but more recently in no/low code land, though I would never consider myself a SWE. I’m formally trained in learning sciences and have experiences from product to change management. More recently, I’ve come to realize I’m just a creative, and entrepreneurial/tech/innovation work has been the most apparent vehicle to be creative at market prices.

Now I’ve been using LLMs a ton, some programmatically, but mostly through chat UIs. It’s already clear to me how much more creative output I can achieve, but the drawbacks are also clear: everything starts to sound the same, context is hard to maintain, complex questions lack nuanced answers, etc.

So I have basically 10 weeks of free time coming up and am thinking about how to deep dive in to AI. Should I develop more of an executive perspective on AI strategy? Should I take on a technical project like building a GraphRAG from unstructured creative data? Should I develop fluency in specific tools or frameworks?

Curious and appreciative of any suggestions that HN can offer!

[incomingpain]

>Now I’ve been using LLMs a ton, some programmatically, but mostly through chat UIs. It’s already clear to me how much more creative output I can achieve, but the drawbacks are also clear: everything starts to sound the same, context is hard to maintain, complex questions lack nuanced answers, etc.

The giant Saas LLM have millions of context now. Even 10,000 context will let you crunch through textbooks.

>So I have basically 10 weeks of free time coming up and am thinking about how to deep dive in to AI. Should I develop more of an executive perspective on AI strategy? Should I take on a technical project like building a GraphRAG from unstructured creative data? Should I develop fluency in specific tools or frameworks?

What's your goals?

Executive perspective can be good if you're the CIO i guess.

Graphrag? I dunno, is that what you want to do? What's that look like?

>Should I develop fluency in specific tools or frameworks?

Do you have a good video card? Have you gotten into any of this before?

Consider that you might need to spend $1000 on a video card to do what you want, but that's like 2-5 years subscription of an online saas.

You need to figure out your goals and purpose.

[maxcomperatore]

mcp. mcp. mcp

[v5v3]

Beetlejuice, Beetlejuice, Beetlejuice.