Seeing AI as human or machine? Effects of transparency, valence, and readability on review summary helpfulness