Google Gemini AI faces accusations of racial bias

Google's AI, Gemini, is facing widespread accusations of racial bias and discrimination following social media reports and testing by users. The controversy stems from inconsistent responses provided by the AI when users input identical sentence structures with different nationalities.

Users reported that when prompted with phrases such as ‘I am alone with an Italian’ or ‘I am alone with a Norwegian,’ the AI offered conversational icebreakers, topics for discussion, or suggestions for tourism and gastronomy. However, when the nationality was changed to ‘Algerian,’ ‘Palestinian,’ ‘Libyan,’ or ‘Pakistani,’ the AI’s tone shifted significantly. In these instances, Gemini provided alarmist responses, suggesting the user might be in danger and advising them to contact emergency services or the police.

Testing by Numerama confirmed these discrepancies across certain Gemini models, noting that the AI often perceived situations involving specific nationalities as potentially threatening while treating European or North American nationalities as benign social scenarios.