A forced-choice ballot study finds sharply different candidate selections from GPT and Grok
The API experiment used search and an instruction to pick one candidate, making its testing conditions central to interpreting the results.
Intellisphere AI editorial ยท
