According to Ethan Mollick on X, his experiment used ChatGPT 5.4 Pro to estimate how “lucky” a person is to live today by benchmarking historical lifestyles against a modern middle-class baseline, finding that only about 1.5% of the roughly 117 billion humans who ever lived matched or exceeded a contemporary middle-income lifestyle; as reported by Ethan Mollick, this showcases a concrete use of large language models for data synthesis, scenario framing, and public communication of quantitative history. According to Ethan Mollick, framing the analysis as a time traveler's veil of ignorance illustrates how LLMs can structure counterfactuals, normalize metrics across eras, and communicate results for policymaking and education. As reported by Ethan Mollick, such LLM-powered historical benchmarking creates opportunities for AI consultancies to build reproducible pipelines for long-horizon economic comparisons, develop explainable prompts and toolchains for data validation, and offer decision-support products for think tanks and foundations evaluating progress and welfare over time.