Winter Rant

Millions of people have contributed and curated knowledge and code in places like Wikipedia and GitHub.

And they have done this for free, in the name of open source.

And now, companies like OpenAI get to just use that open data to train machine learning models and charge money for those AI models?

Here is the bit that is really antithetical to the idea of open source: A lot of these AI/ML tools are so complex, incomprehensible, and resource hungry that they are only available to a select group of individuals and organizations, because only they have the collective brainpower and server capacities to make these systems work.

Open-source contributions – like that on Wikipedia and GitHub – happened with a sense of optimism

… that open source was going to be a way for people to help each other

… that it was going to lower barriers-to-entry for knowledge and technology

… that it was going to get us out of the walled gardens of corporations and paid publications.

And it did! Allow me to play out the metaphor here… For a good decade (or even two!) open-source unlocked knowledge and technology from the technological walled gardens of Big Tech and Media. For the first time knowledge was available at little-to-no cost, in an easy to access manner for anyone with a computer and an internet connection. Soon that access broadened to anyone with a mobile phone and an internet connection. Anyone from anywhere in the world could quickly get up-to-speed on any topic of interest and actually build stuff with shared code from places like GitHub, while having the ability to share their knowledge and learnings back to open source.

But here we are. Back to the walled gardens – this time of OpenAI. Where all that open knowledge has been funneled to machine learning models that are so big that they can only be run using well-resourced server farms. And so, we are back to all of this being a technological walled garden.

Let alone that these AI systems are going to wreak havoc on issues ranging from Fake News to auto-generated essay answers on homework assignments.

But what bugs me is how, once again, a select few individuals, platforms and companies are going to be able to control these AI systems. Just like in the early days of the compiler, the operating system, and the computer network. In each of those technology spaces, we needed an open-source movement to break down those walled gardens. It is hard to imagine what the open-source movement might look like in an AI-heavy technology landscape where the system won’t fit on a commodity laptop.

If current trends continue, it stands to reason that open source is enabling the very walled gardens in technology that it once set out to breakdown.

– vijay, musing about turning a day older.