Text this: Large language models provide discordant information compared to ophthalmology guidelines