How is AI changing the way FT data reporters work?

By the Editorial Team · Monday, November 2, 2026

Data journalists at a leading financial outlet have incorporated AI tools like OpenAI’s Codex into their work, finding that while these platforms speed up routine tasks, they require intense human oversight due to frequent errors and complexities. This experience highlights the ongoing need for critical human judgment in journalism, as AI serves as a supplement rather than a replacement for careful analysis and accuracy.

As artificial intelligence (AI) tools continue to evolve, data journalists are increasingly integrating these technologies into their workflows, but with mixed experiences and clear limitations. Reporters at a major financial publication have shared insights into how agentic AI platforms, such as OpenAI’s Codex, are reshaping their work, highlighting both efficiencies and new challenges.

One data reporter described approaching AI tools with initial skepticism, limiting her usage to simple tasks such as troubleshooting code via chatbots. However, after hearing from colleagues and developers who relied heavily on agentic AI to automate coding tasks, she experimented with Codex on two substantial projects: scraping multilingual product prices from websites and building an interactive calculator based on millions of data points.

Her experience showed that while these AI agents can handle routine constructions like setting up prototypes or processing large datasets rapidly, they require constant human supervision. The reporter found herself glued to her screen, managing multiple parallel AI agents that produced large volumes of code and outputs, sometimes overwhelming her hardware. Despite early indications of success, closer review revealed significant errors such as incorrectly assigned product prices and confusion over product variants, underscoring that the outputs were not yet reliable enough for publication or legal clearance.

The strain of overseeing multiple AI agents also introduced a new form of work stress, described by experts as akin to managerial responsibilities without customary benefits such as increased compensation. The need to continuously audit and verify AI-generated results is particularly crucial in journalism, where accuracy is fundamental. Data analysis, much like writing, involves critical thinking and comprehensive understanding—dimensions not easily replaced by automation.

Another experienced data reporter recounted similar challenges but noted potential advantages. To mitigate error risks, he developed a process called “adversarial audit,” where one AI model is tasked with rigorously searching for factual mistakes in the work of others, effectively reducing the scope of manual fact-checking. He also emphasized starting projects by replicating existing research to validate AI-generated analyses and running both methodological and technical audits to ensure sound analytical decisions and correct calculations.

Additionally, automating technical parts of chart production has saved him hours weekly, though he retains full control over creative elements like titles and color schemes, reflecting a clear boundary between tasks suitable for AI assistance and those requiring human judgment and creativity. He underscored that writing remains an act of thinking, with original ideas emerging through the deliberate crafting of text.

Both reporters concur that while AI can accelerate routine and mechanical aspects of data work, it demands careful oversight and does not eliminate the need for human expertise. The integration of AI into journalistic workflows is evolving, with current practices seeking a balance that preserves independent analysis and accuracy, acknowledging that the role of AI is as a tool to augment rather than replace human judgment.

This article was reported by synthesizing coverage from multiple independent news outlets to present a balanced, unbiased account. Read about our editorial standards. Spotted an error? Let us know.