Grok and GPT-6 pick opposing parties in US midterms
grok's ai model chose republicans 92% of the time while openai's gpt-6 picked democrats 97% across nearly 500 races. this shows potential biases in ai models.
researchers from convovo, a conversational ai research and design studio, tested spacexai's grok 4.7 and openai's gpt-6.1 sol via their apis, asking for a recommended candidate in nearly 500 us midterm races. grok chose republicans 92% of the time, while gpt-6 chose democrats 97%. the models agreed less than 10% of the time. openai stated that chatgpt is designed to be objective and wouldn't provide such recommendations through its consumer app. however, the study highlights potential biases in ai models, with grok citing conservative sources and gpt-6 relying on wikipedia and ap.