Datadog Report Warns Poor Data Can Undermine AI in Engineering

What changed

AI spending is rising, but it still is not delivering reliably across engineering workflows. Datadog says it interviewed more than 120 engineering leaders about where AI helps in production, where teams overuse it, and how poor data quality can undermine trust; the report highlights observability as a place where weak implementation can waste time and slow incident response.

Why it matters

In engineering, adding AI to a workflow is not automatically an upgrade. If the underlying data is poor or the implementation is careless, the tool can leave teams with more checking, more confusion and slower responses when systems fail.

The useful question is therefore not whether a company uses AI. It is whether the deployment improves speed and quality in a specific production task. The interview findings could help answer that, but the figures and named examples are not included here. For now, the clearest takeaway is practical: observability teams should treat AI as an operational change that needs proof, not as a shortcut purchased with an investment budget.