Which is the better data analyst? Benchmarking ChatGPT vs. Claude (www.geckoboard.com)

A recent benchmarking exercise compared the data analysis capabilities of AI assistants Claude and ChatGPT, assessing their performance on realistic customer support tasks using a shared live dataset of Zendesk tickets. The two models were put through a seven-stage analysis process that included orientation, performance reviews, focused investigations, visual presentations, and self-audits. While both models demonstrated high accuracy in retrieving data and did not miscalculate numbers, they struggled with framing and context, sometimes creating misleading interpretations based on the data presented. The results revealed that Claude slightly edged out ChatGPT in shared analytical scores, particularly post-audit, with scores of 91.3% compared to ChatGPT's 83.7%. Notably, Claude exhibited a willingness to reassess its findings, often self-correcting its initial assumptions, while ChatGPT took a steadier, more procedural approach. Both AI models showcased strengths in speed and communication but highlighted the importance of structured processes and explicit prompts to avoid common pitfalls in data analysis. This comparison underscores the growing significance of AI tools in knowledge work and prompts further exploration into how these systems can better emulate the rigor of human analysts in interpreting data.