48 APPENDIX CONT. Exhibit 1.3: Generative AI Efficiency Increase Calculations. Described in Steps 4-6 in Exhibit 1.1. Representative Occupational Category Tagged Workers per Category Applicable Efficiency Gain [3] Base Hours of Ouput per Man- Year Per Person [1] Base Hours of Output per Occupational category AI-Improved Hours Output Per Man- Year per Person [2] AI-Improved Outputs per Occupational Category Basic Office Support and Operations 132,220 14.0% 2080 275,017,600 2371.2 313,520,064.0 Intermediate Office Operations 69,612 40.0% 2080 144,792,960 2912 202,710,144.0 Technical Tasking and Specialized Roles 26,322 55.8% 2080 54,749,760 3240.64 85,300,126.1 Field-based Operational Work 824,666 6.5% 2080 1,715,305,280 2215.2 1,826,800,123.2 Administration/Knowled ge-based Professions 849,046 25.1% 2080 1,766,015,680 2602.08 2,209,285,615.7 Total Workforce Output Increase (Hours of Output per Year) 681,734,793.0 Total Workforce Output Increase (% Increase) 17.2% AI Efficiency Increase Calculation [1] Establishes baseline hours per man-year as 40 hours per work week, 52 weeks per year [2] Assumes increased hours output per man-year compared to baseline [3] While the Goh, Gallo, and Strong et al. study notes an increase in average case time with the use of GPT4, we assume that this is due to short-term inefficiencies associated with user adoption, given the nascency of LLM tools. As individual users gain experience, in the long-run, we believe reductions in error rates may be the lasting factor causing improvement in the speed of work streams due to less rework for cases. Generative AI Adoption in the US Military 50 Brynjolfsson et al. (2023). Generative AI at work (NBER Working Paper No. 31161). National Bureau of Economic Research. https://doi.org/10.3386/w31161 51 Noy and Zhang, (2023). Experimental evidence on the productivity effects of generative artificial intelligence. Science, 381(6654), 187–192. https://doi.org/10.1126/ science.adh2586 52 Peng et al. (2023). The impact of AI on developer productivity: Evidence from GitHub Copilot. arXiv preprint arXiv:2302.06590. https://arxiv.org/abs/2302.06590 53 Goh et al. (2025). GPT-4 assistance for improvement of physician performance on patient care tasks: a randomized controlled trial (Nat Med 31, 1233–1238). Nature Medicine. https://www.nature.com/articles/s41591-024-03456-y 54 Dell'Acqua et al. (2023). Navigating the jagged technological frontier: Field experimental evidence of the effects of AI on knowledge worker productivity and quality (HBS Working Paper No. 24-013). Harvard Business School. https://www.hbs.edu/faculty/Pages/item.aspx?num=64700