Chromatics AI report finds models exclude non-Western languages and perspectives
Chromatics AI released a report pointing out how many AI systems are trained on data that does not adequately represent the languages, worldviews, and communication styles of large parts of the global population, which leaves out other cultures.
This matters because AI is now used for things like education and legal advice, and if the tech doesn't reflect everyone, it can miss the mark for people outside the West.
Translation alone overlooks cultural differences
The report highlights that simply translating apps isn't enough; deeper cultural differences still get overlooked.
Amazon had to retrain Alexa with local data for Mexico, showing how tricky it is to make AI truly inclusive.
There are cool efforts worldwide, like Nigeria's N-ATLaS and Thailand's OpenThaiGPT, to create AI that respects local languages and cultures.
Chromatics AI's founder Larry Adams says we need community-driven datasets so AI works better for everyone.