Only 37% of Developers Trust AI for Incident Response. The Gap Is Solvable.
Cloud Platform EngineeringIT leaders and the developers who actually own on-call rotations disagree sharply about AI’s readiness for incident response. According to PagerDuty’s State of AI-First Operations survey, 59% of IT decision-makers expect AI to improve downtime and recovery by more than 20%. The share of developers who agree: 37%. That 22-point gap isn’t just a perception problem — it’s the difference between a platform investment that gets adopted and one that doesn’t.
Context
The business pressure on incident response has never been higher. PagerDuty’s survey found that 68% of organizations lose more than $300,000 per hour during IT incidents, and IT uptime has become a board-level priority in that environment. The response from platform teams and vendors has been predictable: push AI-assisted triage, automated root cause analysis, and ML-driven alerting into the incident workflow. The assumption is that developers will welcome anything that reduces on-call burden.
The survey data complicates that assumption. Forty-two percent of operations leaders acknowledge that outages contribute to developer burnout — but the same developers who are burning out are also the ones who don’t trust the AI tools being handed to them. The tooling being deployed to solve the burnout problem is not being trusted by the people it’s meant to help.
The Detail
The 22-point trust gap has a structural explanation: developers interact with AI in incident response at the moment when the cost of a bad recommendation is highest. An AI-generated root cause suggestion during a live P1 incident is not an intellectual curiosity — it’s a decision input during high-stress, time-constrained triage. Developers have accumulated experience of AI tools that are confident and wrong; incident response is not the context where “probably right most of the time” is acceptable.
IT decision-makers, who are farther from the incident console, are evaluating AI on aggregate performance metrics and strategic potential. Developers are evaluating it on the specific experience of the last time the AI pointed them in the wrong direction at 2am. Those are different evaluation frames, and the 22-point gap reflects them accurately.
The workforce implications compound this. PagerDuty’s survey found that 51% of organizations expect to hire or reskill staff specifically for AI-powered incident detection. That’s a significant investment signal — but it also implies that the current workforce isn’t yet positioned to use these tools effectively. The skill gap is real on both sides of the interface: the tooling needs to be more trustworthy, and the teams using it need more structured exposure to how it fails before they can rely on it where it works.
The So What
Platform teams deploying AI in incident response workflows face a sequencing problem: the business case for AI in on-call is strong, but developer adoption requires trust, and trust requires demonstrated reliability in lower-stakes contexts before high-stakes deployment. The organizations that close the gap fastest will be those that give developers visible, audited exposure to AI recommendations before those recommendations become autonomous actions — starting with noise reduction and alert correlation, building toward root cause suggestion, and reserving autonomous remediation for categories of incidents where the AI’s failure modes are well-characterized. Deployment velocity matters less than building the trust that makes developers actually use what’s deployed.
Photo by hitesh choudhary on Pexels
Source: The New Stack
Content created with AI assistance and reviewed for accuracy.
Join the conversation
Stack Insiders is our free community for readers who want to go deeper — share resources, ask questions, and connect with others across every vertical we cover.
Join Stack Insiders →