AI companies are becoming the economists of their own impact

The Financial Times examines technology companies’ expanding role in economic research. AI firms can observe usage that public researchers cannot, but control over data also influences the questions researchers can answer. The issue is broader than whether an individual economist produces favourable conclusions: access itself shapes the research agenda.

Ramp’s June study illustrates both the value and the limits of company-held evidence. Combining spending records with employment data for more than 21,000 firms, it found that intensive AI adopters increased headcount by about 10% over two years, with entry-level employment rising about 12%. Low-intensity adopters showed no statistically significant overall employment change. Those figures complicate a simple account in which buying AI automatically means eliminating jobs.

They do not establish that AI caused the hiring. Ramp also found that adopting firms were already more likely to be fast-growing, venture-backed and engineering-intensive. Such businesses may have hired more without the technology; aggregate headcount can also conceal changes in occupations, workload or recruitment standards. This June evidence provides context for the FT’s October discussion, not a new October result.

The useful standard for this research is whether an outsider can understand the sample, exclusions, adoption measure and alternative explanations. Publicly documenting those choices—and allowing scrutiny of unwelcome findings—makes proprietary data more informative. A large dataset alone cannot distinguish a technology’s effects from the characteristics of the companies most eager to buy it.