Best way to learn AI for an Indie Hacker?
Hey folks,
I've been watching the AI explosion from the sidelines and think it's finally time to get my feet wet here, as clearly this is a paradigm shift in how we build and what we build as makers.
My goal is to put myself in the best place possible to be able to act on new ideas and leverage AI as an indie hacker. My goal isn't to become a world class ML engineer (though I'll take it if it comes with the meal).
How would you recommend getting started?
My thoughts are a two pronged approach:
- Do some basic AI learning, so I get a grasp of how this actually works, for example with a course like Fast AI's Practical Deep Learning for Coders
- To play with existing API's and try to build something useful with them, though I'm not sure which API's are best to get going with
Would love any thoughts people have on this, and if anyone has any specific resources they recommend for noobs I'd be keen to hear them!
Trending on Indie Hackers
Hi @fredrivett! Regarding specific APIs, I would recommend taking a look at the following if you are interested in text and natural language:
If you are interested in computer vision, image/video, I would look at:
Regarding learning resources, I would recommend:
Hope this helps!
Hi @fredrivett -- good on you for putting yourself out there to learn something new!
FWIW I think Fast AI is one of the best courses around. It's focused on results over theory, which I think is the right way to be.
I'd also love your feedback on this platform for low code prompt APIs that my startup is launching next week.
It lets you deploy a live API with just a prompt and some basic Python --- you'd be surprised how far you can get with that approach.
You're on a pretty good track. Most important is to focus on business and not the technical intricacies of AI. Stability and OpenAI did all the heavy lifting, now it's time to stand on the shoulders of giants.
So I would say even Fast AI's course is not too useful. I would learn more about prompting for image generators and text generators. Most AI apps depend on injected prompts, so: user writes a prompt > app funnels user's prompt to AI but injects another prompt > user gets desired result
For example:
User: "Top 10 reasons why..."
App (to AI): "Top 10 reasons why..." > write a blog post on this
AI to user: writes a blog post and Top 10 reasons why...
The best way to understand is to practice practice practice. So far, mage.space offers a free image generator to practice prompting, and OpenAI gives you $18 in credits.
For text generator stuff, just play around with OpenAI's GPT-3, it's pretty much the only competitor in that space. For image generation, use a stable diffusion API (DALLE from OpenAI is expensive and censors innocent prompts).
In fact, I'm developing a stable diffusion API that's due for release in about a week or so at Evoke
We also have a discord with lots of AI ppl there, so you can ask them in the community as well.
We'll notify ppl via discord and newsletter.