Quantifying truth and authenticity in AI-assisted candidate evaluation: A multi-domain pilot analysis