According to @emollick on Twitter, a new NBER working paper (w34851) has been released, and according to the National Bureau of Economic Research (NBER), the paper provides empirical evidence on how generative AI tools impact knowledge worker productivity, task quality, and adoption patterns. According to the NBER paper, results highlight measurable efficiency gains on complex writing and analysis tasks when workers use large language models, with the largest improvements among lower baseline performers, indicating potential skill compression effects. As reported by NBER, the study also documents shifts in task allocation and complementarity with human judgment, suggesting that firms can realize near-term ROI by targeting workflows such as drafting, customer support, and data summarization while instituting guardrails for accuracy and oversight. According to NBER, the paper discusses organizational implications including changes in training, evaluation, and IT procurement, and outlines business opportunities in AI copilots, domain-tuned models, and workflow orchestration that reduce time-to-value in enterprise settings.