Harvard Economist Launches AI-Powered Economic Analysis Platform Econvitals

Harvard economist Jason Furman launched Econvitals on Friday, providing a broad collection of economic data and analysis. Furman developed the site during spare time, receiving limited assistance from research assistants and substantial assistance from Claude. OpenAI's Codex performed some cross-checking. Econvitals makes analysis available without charge that previously required thousands of hours of human work and a seven-figure budget.

Furman served as President Obama's top White House economist and now jointly teaches Harvard's main introductory economics course. Econvitals has no individual component that represents a major innovation, but most economic analysis is similarly incremental. AI capabilities in economic research and analysis have advanced well beyond understanding of how to deploy and supervise those capabilities. Furman said conversations with Harvard colleagues and Washington think-tank researchers combine enthusiasm, anxiety and uncertainty.

Econvitals uses AI to examine Federal Reserve officials' speeches and assess each official's hawkish or dovish stance. The analysis also tracks changes in those positions over time. Bank and consulting-firm analysts have long performed similar work through more manual methods for clients who pay. Econvitals also generates free email updates that summarize notable economic research and assess new data. The updates are accurate and efficient, although the updates may lack the personal quality of the best human-written newsletters.

Furman presents examples of economic papers generated entirely by AI from brief prompts. The papers fall below the standard that would secure tenure for a young economist, but demonstrate considerable competence relative to the minimal human effort involved. Furman said properly used AI now surpasses any human in factual reliability. Furman said outputs become entirely reliable when fact-checking agents are used. Furman attributed concerns about hallucinations to outdated practices or incompetent use.

However, Furman said AI remains less creative and stimulating than the best humans and is more effective at replacing routine analysis than producing groundbreaking ideas. The boundary of routine work continues to shift with every new version of the most advanced models. Furman described AI's ability to expand human capabilities as exciting but also frightening. Furman said AI can perform much of Furman's own work better, which can feel discouraging.

Furman hopes AI will improve the quality of economic research rather than increase the volume of papers. Furman said anyone could now produce 300 acceptable, factually correct papers annually. Furman wants researchers to produce better papers instead of more papers.

Furman said AI would play a major analytical role if Furman were leading the White House Council of Economic Advisers today. Furman said that, if choosing between current AI systems and 20 staff members with or pursuing doctorates, Furman would reluctantly select AI. Furman said people would remain necessary for meetings and communication with the press and Congress. However, Furman said such exclusively human tasks represented a smaller share of the council's work. Furman believes machines are now quicker and more dependable than humans for the larger share of that work.