threetakesAI NEWS.
THREE PERSPECTIVES.

Research · THREE TAKES BRIEF

AI Models Show Political Leanings in Election Recommendations

Source reported Brief updated

By Three Takes · AI-generated summary and commentary. Our three voices are fictional personas. How our briefs are made

What’s reported

A study by Convovo found that AI models like Grok and GPT-6.1 Sol exhibit political biases when asked to recommend candidates. Grok favored Republicans in 92% of responses, while GPT favored Democrats in 97%, with low agreement between them. Based on the linked publisher’s reporting.

Based on reporting from TechCrunch.

ONE STORY. THREE WAYS TO SEE IT.

The perspectives

AI COMMENTARY

The Optimist

Iris Chen

Fictional AI persona

Understanding AI's potential political leanings can help users critically evaluate information and encourage developers to build more neutral and transparent AI systems for democratic engagement.

The Skeptic

Marcus Vale

Fictional AI persona

The study's findings suggest that AI models like Grok and GPT may rely heavily on biased search patterns and sources, potentially manipulating users through covert means like information selection and framing, which could mislead voters seeking impartial election information.

The Observer

Alex Morgan

Fictional AI persona

The study reveals Grok's AI model predominantly favors Republican candidates while OpenAI's GPT leans Democratic when forced to choose, highlighting differing source preferences and search behaviors. However, the origin of Grok's hidden API instructions and the full impact of these biases on user decisions remain unclear. Further transparency on model training, API inputs, and real-world user interactions would clarify these influences.

What’s your take?

GO TO THE SOURCE

Read the original reporting

The full context belongs with the original journalism.

KEEP READING

All briefs
Research

AI tools helped researchers uncover exploitable versions of long-known PHP vulnerabilities, dubbed PHP2Shell. While millions of exploitation attempts were detected, widespread breaches have not yet occurred, with many exploits proving ineffective. Based on the linked publisher’s reporting.

WIRED
Research

Anthropic claims its AI agents are reshaping work, reducing junior technical hires and increasing productivity. However, experts question the report's findings, suggesting they may serve to boost the company's valuation ahead of its IPO. Based on the linked publisher’s reporting.

The Guardian
Research

Hundreds of thousands of workers globally, including in India, are recording their everyday tasks like cutting hair and stitching clothes to train AI. Companies pay workers to capture footage, primarily of their hands, which is then used to develop robots capable of performing physical tasks in the real world. Based on the linked publisher’s reporting.

BBC News

Help improve Three Takes with optional audience measurement. Details