The authors use O*NET data and a lookup table of human abilities affected by AI (e.g. oral comprehension; inductive reasoning; etc) to estimate the impact of generative AI across occupations and demographics. They find that the greatest impact will likely be on highly educated, highly paid, highly creative occupations.
The specifics of their findings are broadly consistent with Briggs et al. - 2023 - Large Effects of Artificial Intelligence on Economic Growth. One significant difference I notice is that they are much more pessimistic about teachers of various kinds.