Yandex Trains AI to Better Understand Cultural Cues in Image Generation

Yandex’s image-generation neural network is getting a cultural upgrade. The upcoming update will help the AI better interpret national identity in user prompts.

More than 8 million users have adopted Yandex’s 'Shedevroom' AI to create images, but the tool has occasionally struggled with cultural accuracy. In its latest update, the company is retraining the model to better understand and reflect cultural nuances.

The move comes after users flagged errors in responses to prompts involving national identity. In one case, the AI generated an image of an astronaut with an American flag in response to a request for a 'Russian cosmonaut.'

Yandex attributes these issues to the open datasets on which the model was trained and is encouraging users to continue reporting inaccuracies. The company told RIA Novosti that ongoing algorithm refinements aim to boost the cultural relevance of generated content.

The forthcoming update is expected to significantly improve the AI’s ability to match image output with the intended cultural context.