How the New York Times Uses A.I. For Journalism (www.nytimes.com)

The New York Times says it is increasingly applying AI and machine‑learning tools across newsroom workflows — but not to write articles. Practical uses include computer‑vision models that sift satellite imagery to flag likely bomb craters for human investigators, recommender systems that personalize homepage and article suggestions based on reading history, rough location and popularity signals, and generative AI to draft headlines, summaries and other short editorial text. All uses are coupled with training, editorial guidance and mandatory human review; journalists retain final responsibility for everything published. For the AI/ML community this underscores a real‑world, hybrid model of deployment: models accelerate data triage, discovery and distribution while humans provide verification, editorial judgment and ethical oversight. Key technical implications are clear — applied computer vision for large‑scale imagery analysis, recommender algorithms leveraging behavioral and geolocation features, and the controlled use of generative models for auxiliary copy — each requiring accuracy, bias mitigation, provenance tracking and explainability. The Times’ approach highlights opportunities to scale investigative work and personalization, but also the necessity of guardrails, transparency and workflows that prevent automation from supplanting journalistic accountability.