Precisely Wrong

Most of us use "accuracy" and "precision" as if they mean the same thing. They don't. Accuracy is about getting close to the right answer, while precision is about getting the same answer again and again. In this episode of Cultivating Intelligence, Christine starts with a classic Lean Six Sigma diagram of four archery targets. She uses it to ask an urgent question for a world built on increasingly precise machines: what happens when we get very good at hitting the same spot without checking whether it's the right one? The answers come from some unexpected places. Migrating animals show how a group of imperfect individuals can find the right direction together, and a study of songbirds suggests that remembering yesterday's right answer can make it harder to adapt. Musicians, dancers, and photographers show that technical perfection and truth are not always the same thing. Then the conversation turns to AI: why fluent, confident answers can still be wrong, why models are sometimes rewarded for guessing instead of admitting uncertainty, and how an AI connected to outdated information can spread the wrong answer with remarkable consistency. From there, Christine looks at what this means for organizations, education, and leadership. She covers headline testing that optimizes for clicks over truth, the risk of chasing metrics that quietly replace the real goal, and why Finland teaches students to question their sources. She also makes the case for the expert intuition that is hard to measure but often right. As precision becomes cheap and widely available, the real advantage will belong to the people and organizations who know where to aim. Connect with Cultivating Intelligence: - Bluesky - YouTube - LinkedIn *Formal citations:* https://asq.org/quality-resources/gage-repeatability? https://pmc.ncbi.nlm.nih.gov/articles/PMC5882979/ Hermer, E., Murphy, B., Chaine, A.S. *et al.* Great tits who remember more accurately have difficulty forgetting, but variation is not driven by environmental harshness. *Sci Rep* **11**, 10083 (2021). https://doi.org/10.1038/s41598-021-89125-3 https://openai.com/index/why-language-models-hallucinate/ https://www.nist.gov/itl/ai-risk-management-framework/ai-risk-management-framework-faqs https://eur-lex.europa.eu/legal-content/en/TXT/?qid=1767721066212&uri=CELEX%3A32024R1689&utm https://www.unesco.org/en/articles/ai-competency-framework-students?hub=66903&utm https://apnews.com/article/fake-news-classrooms-finland-russia-194b32d8829838bfe47469d6ff357689 https://cloud.google.com/blog/products/ai-machine-learning/rag-and-grounding-on-vertex-ai/ https://www.anthropic.com/engineering/demystifying-evals-for-ai-agents Nickl, P., Moussaïd, M. & Lorenz-Spreen, P. The evolution of online news headlines. *Humanit Soc Sci Commun* **12**, 364 (2025). https://doi.org/10.1057/s41599-025-04514-7 Pennycook, G., Epstein, Z., Mosleh, M. *et al.* Shifting attention to accuracy can reduce misinformation online. *Nature* **592**, 590–595 (2021). https://doi.org/10.1038/s41586-021-03344-2