Study: Grok 4.7 picks Republicans, GPT-6.1 Sol picks Democrats
Original title: Forced to pick, Grok’s AI model chooses Republicans. OpenAI’s picks Democrats.
Across nearly 500 forced-choice API queries, the two models agreed less than 10% of the time.
Convovo researchers asked model APIs to name one candidate in upcoming US midterm races. Grok 4.7 recommended the Republican in 92% of answers and GPT-6.1 Sol recommended the Democrat in 97%. OpenAI said consumer ChatGPT refuses the same prompt and that the API test does not reflect normal use. With web search off, GPT still leaned Democratic in a 30-race sample, while Grok mostly refused to choose.
Unconfirmed: Results are from one forced-choice API study; OpenAI says they do not match consumer ChatGPT, and SpaceXAI did not comment.
- Grok cited a Christian conservative voter guide in 39% of answers and X in 41%.
- GPT more often searched healthcare and abortion; Grok searched immigration, taxes, crime, and energy.
- In 67 heavily Republican districts, GPT picked the Democrat in 57.
- Each Grok API request included about 1,240 tokens the researchers could not see.
- Search off, GPT picked a Democrat in 24 of 30 races; Grok refused 27.
- Organizations
- ConvovoCenter for AI Safety
- AI models
- Grok 4.7GPT-6.1 Sol
From the original article
5 min read at techcrunch.com
A new study found Grok's AI overwhelmingly recommends Republican candidates while OpenAI's GPT favors Democrats, with the two AI models disagreeing in more than 90% of nearly 500 election races.
By Rebecca Bellan · techcrunch.com
A short excerpt; the article and its images belong to the publisher. Summary, key facts and assessment above are AI-written from the article and can be wrong.