Why a global nonprofit is building its own version of AI

Current AI is spending millions in funding to help communities build their own AI tools that respect their local needs. They recently partnered with the Indian government to create a small, portable device that works offline and understands 22 different Indian languages. This is part of a larger effort to ensure that AI does not just serve the interests of tech giants. They are also funding groups in places like Kenya, Lebanon, and the Amazon to help locals collect their own data and build digital tools that preserve their history and knowledge on their own terms.

To understand why this matters, think of how an AI learns. These tools are essentially massive digital students that have read almost everything on the internet. They learn by scanning vast amounts of text, which is called training data. If the data is mostly in English, the AI becomes an English expert but remains clumsy or ignorant when it comes to other languages. This is where the problem lies. If a company scrapes the internet to train its AI, it might accidentally include private or sacred cultural information without asking the people who own that knowledge.

We often treat AI like a finished product we just download, but it is actually a reflection of the people who built it and the data they chose to include. If only a few wealthy companies control the technology, they get to decide which perspectives are important and which languages count. A public alternative, like the one Current AI is proposing, treats technology as a shared resource—much like the early internet was supposed to be. If AI is going to shape how we work, learn, and preserve our memories, it is worth asking whether that power should belong to shareholders or if it should be something rooted in the communities it serves.