Open source AI Economics - Blog

Hey Folks, my name is Ashish Singh, I wrote a blog post, I’d like to share with you all.

Ever since Meta introduced Llama models as open-source, I have been intrigued by the idea of open-sourcing AI models. What surprised me most was that it took a humongous amount of money to develop these models and yet Meta open-sourced the model, practically giving up their research and its output for free. I downloaded the model on my M3 MacBook Air and it worked flawlessly. I couldn’t understand why though.

  1. Why would companies (for profit) open-source their cutting-edge research and models?
  2. How does it help their case?

I’ve been asking these questions to anyone and everyone around me on hacker news, on Reddit, etc. to no fruition.

Finally, here I am on my exploration of the economics of open-source AI Large Learning Models (LLMs). Check out the full post

Critique please.

2 Likes

If you see the general history of AI for last decade, last 2 years were an aberration. Opensource has always reigned supreme and any moat with closed source models has always dissipated quickly.

I was explaining to a journalist that a 500W system that you can buy for 2.5 lakhs (JOHNAIC) can now run models which are better than GPT3.5 that set off this closed source anamoly. I am only more excited about this as newer models get better and smaller.

1 Like