Bias in a Dataset – The Bread Problem

Bias in a Dataset – The Bread Problem

by Rabin Puri -
Number of replies: 0

My example: I think international AI systems may misunderstand Nepali foods such as sel-roti, dhindo, gundruk, and traditional local dishes because their datasets may contain more Western food images than Nepali food images. AI may also misunderstand Nepali names, local languages, festivals, and rural classroom practices.

Who may be excluded: This may affect students from rural communities, government schools, and families who speak Tamang, Newari, or other local languages. Their culture, language, and learning needs may not be properly represented in AI-generated information.

How I would explain it: I would tell students that AI learns from examples and information collected from different sources. If it does not get enough examples from our local communities, it may give incorrect or unfair answers. Therefore, we should provide diverse examples, respect every culture, and always check AI-generated information before trusting it.