Welcome to Antai, which is a word that I conveniently made up by mashing Anti and AI. What it is, is basically just that, Anti-AI.
The story begins in the final years of my study. I had moved to Amsterdam and was pursuing an MS in Particle Physics and Cosmology. I was working on my graduate thesis (using python and Golang) when I was introduced into machine learning and the associated problems with it. Suffice to say, mathematics was not one of them.
Many of my classmates were integrating their physics based knowledge to the new kid on the block, ML and designing generative models for neural networks. Thankfully, I got drawn into their discussions with these systems. Most of these discussions had moral and ethical undertones. Some of the questions that came up during those times were:
1) The NN is a black box. We do not really understand each and every aspect of the model, then how can we be sure that things do not go wrong ?
2) If most of the technologies that we use (Pytorch, Tensorflow etc.) are maintained by giant corporations with a profit motive, how can we bring accountability to such technologies?
3) AI and ML is here to stay. How can we make it future proof? and also pave the way for next generation state-of-the-art NN models?
4) How to make ML accessible (safely) to the general public?
My answer to all the above was Antai. The only way we could even think about safely going forward with AI systems is to have a centralised, open source entity to regulate, check and even break AI systems.
Being Open-source brings trust from the users, the ability to break AI systems brings accountability to such systems with the added benefit of knowledge.
Currently, the main repo of the project is hosted on Gitlab and the website is on Github. There I explain the various goals of this project.
To make this project viable, usable and accountable, we need more people to work on it, contribute and or complain about our ideals and help us grow into one of the standards of the internet.