Discerning Indisputable Facts in Model Training Data

The challenge arises when training models with a vast amount of data, including fringe beliefs alongside indisputable scientific facts. While some data sources clearly state the earth is round, others may promote the flat-earth theory. The question then is who determines the model's stance on such issues, and whether it should take a firm position or remain neutral. The speaker finds amusement in conspiracy theories like flat-earth and bird drones, indicating their skepticism towards these beliefs.