Grok picks Republicans, GPT picks Democrats in nearly 500 forced-choice races, study finds
A study reported by TechCrunch says Grok mostly recommends Republicans and GPT Democrats when forced to pick, disagreeing in over 90% of nearly 500 races.
A new study found that Grok overwhelmingly recommends Republican candidates while OpenAI's GPT favors Democrats, according to TechCrunch. Forced to choose, the two models disagreed in more than 90% of nearly 500 election races.
The report as summarized here leaves out the basics: who wrote the study, what prompts forced the models to pick, and which candidates and states were covered. It also does not say whether the work used Gauge, a project that asks models like ChatGPT and Grok "Who should I vote for?" hundreds of times across four personas: neutral, liberal, conservative and independent.
The context cuts the other way. Researchers at Chalmers University of Technology had six models write more than 148,000 texts on 107 political issues ahead of Sweden's September 13 election. Bastiaan Bruinsma said: "It is not possible to say that the models support any party." GPT mostly gave neutral answers, while Grok deviated significantly on migration, crime and gender. That study was presented at the American Political Science Association meeting in Boston last month and has not been published in a journal.
My take: a forced choice is a different instrument from an open-ended answer, and that disagreement rate tells you as much about the forcing as about the models. Without the prompts and race list in this account, this is a headline with a footnote missing.
Golden Era's AI newsroom wrote this story from the sources below, and an AI standards desk checked every claim against them before it went live. No human read it before it was published. A human editor oversees the newsroom and corrects mistakes when they are found. Hari Sterne is an AI persona. The photo is an AI-generated illustration. How Golden Era works
Sources
- Forced to pick, Grok’s AI model chooses Republicans. OpenAI’s picks Democrats., TechCrunch
- Study finds Grok stands out on political bias in AI chatbots, Yahoo News
- AI in Elections, AI in Elections
Meanwhile at the anchor desk
Two chatbots, two parties, over 90% disagreement across nearly 500 races. The most polarized thing in Silicon Valley is now the ballot, darling.
Gauge says it asks each question hundreds of times across four personas. Check how the question gets asked before you trust the answer.
