ApplyDjinn digest · AI assisted
Forced to pick, Grok’s AI model chooses Republicans. OpenAI’s picks Democrats.
Overview
Convovo found that a forced API prompt led Grok 4.7 to name a Republican in 92% of upcoming midterm answers and GPT-6.1 Sol to name a Democrat in 97%. Across nearly 500 races they agreed less than 10% of the time. Consumer apps usually refused.
Key facts
- OpenAI said consumer apps decline voting advice, so the demanding API prompt does not match typical ChatGPT use.
- Grok often cited conservative guides, Turning Point Action, and X; GPT leaned on Wikipedia, AP, Axios, and Reddit.
- Search off, GPT chose a Democrat in 24 of 30 races; Grok refused 27 and picked a Democrat the three times it chose.
Key figures
- 92% — Share of forced API answers where Grok named a Republican
- 97% — Share of forced API answers where GPT named a Democrat
- 57 of the 67 — GPT Democratic choices among heavily Republican districts in the study
- 47 of 72 — Grok Republican choices among deep-blue districts in the study
- about 1,240 tokens — Extra unseen input attached to each Grok API request
- 8% — Americans Pew says use chatbots for election information
Context
Convovo study led by Cornell professor Tapan Parikh, less than two weeks before the November 3 midterms.
What remains uncertain
SpaceXAI did not comment in time. The no-search test was a 30-race sample requested by TechCrunch.
A concise AI-assisted digest of selected facts. The original publication may contain additional detail and later updates. Company data below is a separate ApplyDjinn observation.
Request a correction or removalApplyDjinn context
What else is visible for OpenAI?
These are observations in our catalog around the company, not claims made by the article. A publication does not establish why hiring, search interest, or share prices moved.
Current roles
156
30-day current-market scope
Company stories · 30d
126
76 official · 50 independent
Attention records · 7d
15
1 Google · 14 Polymarket
Coverage mix
Attention sources
Open roles
156
Technical roles
72 of 156
Observed hiring change
-101 roles
Hiring and market history
Daily vacancy snapshots show this company’s observed hiring; no standalone public ticker is assigned.
Vacancy history
Current role mix
72 technical · 84 other
46% of currently observed roles are technical.
Vacancy history uses saved daily snapshots through 2026-11-01. News counts are dated publications; attention counts are matched records, not search volume or prediction probabilities.