ALT Fig 1. Schematic illustration of how “PLOS LLM” could interact with humans, considering both knowledge generation and retrieval/synthesis processes. Text in blue illustrates how new research/knowledge generation could interface with PLOS-LLM, and the steps in orange illustrate how knowledge retrieval/synthesis could work, for example allowing users to ‘zoom in and out’ from looking at headline answers to a question, all the way down to exploring the datasets of individual studies.

ALT Fig 1. Schematic illustration of how “PLOS LLM” could interact with humans, considering both knowledge generation and retrieval/synthesis processes. Text in blue illustrates how new research/knowledge generation could interface with PLOS-LLM, and the steps in orange illustrate how knowledge retrieval/synthesis could work, for example allowing users to ‘zoom in and out’ from looking at headline answers to a question, all the way down to exploring the datasets of individual studies.

ALT Fig 1. Study design and methods. * Of 715 ineligible individuals, 343 were ineligible for more than one reason. Reasons for ineligibility included: not ≥18 years old (n = 2), not cisgender or transgender male (n = 79), does not have sex with cisgender or transgender males (n = 37), not HIV-negative (n = 101), not on PrEP (n = 367), not sexually active in the last three months (n = 67), and CAGE-AID score <2 (n = 562). ** Of 18 participants who did not pass all validity checks, 1 participant failed to pass more than one validity check. Reasons for not passing all validity checks included: age and date of birth did not match (n = 15), home zip code and home state did not match (n = 2), and IP address did not confirm current location in the US (n = 2).

ALT Fig 1. Study design and methods. * Of 715 ineligible individuals, 343 were ineligible for more than one reason. Reasons for ineligibility included: not ≥18 years old (n = 2), not cisgender or transgender male (n = 79), does not have sex with cisgender or transgender males (n = 37), not HIV-negative (n = 101), not on PrEP (n = 367), not sexually active in the last three months (n = 67), and CAGE-AID score <2 (n = 562). ** Of 18 participants who did not pass all validity checks, 1 participant failed to pass more than one validity check. Reasons for not passing all validity checks included: age and date of birth did not match (n = 15), home zip code and home state did not match (n = 2), and IP address did not confirm current location in the US (n = 2).