the journey of a thousand miles begins with a one-step
- a Chinese proverb
AI has always been a topic of fascination for everybody. We have always wondered about the possibilities of creating intelligence ourselves - putting us in the realm of gods. It always seemed out of our reach but the latest innovation around the field of Deep Learning, Large Language Models has brought us closer than ever before. models like Chatgpt and stable-diffusion have taken the world by storm and shown us how powerful this tech already is, and the pace just keeps increasing.
Open-source AI and Closed-source AI
AI is going to be a fundamental shift to how we do work - very much like the internet. However lately we are seeing a split on how AI is being built.
Huge companies with massive resources are building these models and keeping them behind closed doors. Take for example the latest OpenAI’s chatgpt model. Everyone can interact with the model via the UI or use the APIs to build apps on top of it but OpenAI does not share the model weights. This means you have no freedom to modify the weights or make any changes or improvements to them.
On the other spectrum of this was stable diffusion. When stable diffusion was released, they shared the weights, dataset, code and paper which gave the tools for people to build on top of and the rate of innovation from that has been mind-boggling. People build all sorts of stuff on top of it and made rapid improvements to the model, reducing the model size so that it is accessible to more people, and adding more capabilities like in-painting, personal avatar design and on and on. This was a testimony to the rate of innovation that can happen if it is built out in the open.
But what does all this have to do with this blog?
There is however a lot of stuff to figure out with Open source AI like whether it would be possible to repeat what stable diffusion did, on the other hand, ML models need huge amounts of compute, how will we find that and pay for it in the long term? What are the unique advantages open-source AI brings to the market (because in our capitalistic society that is all that matter)?, is that enough for it to win over closed-source AI? how would both of them play out? will they exist together? or will one take over the other?
🎯 These are some of the questions we want to find the answers to and we’ll be sharing them as we explore these ourselves in this blog
We believe Open-source AI is the future and the goal is to explore and build that future. Together, We are going to learn about the important machine learning concepts behind them, explore organisations and projects that are building in the open and figure out how we can contribute to these projects to make our vision a reality.
So keep an eye on this space if these topics excites you and we will see you in the next post 🤞🏻.