AI-Driven Incident Management Needs Human Empathy

Forbes

10-07-2025

Artificial Intelligence
Artificial intelligence (AI) is becoming woven into the fabric of our lives. In fact, according to a recent Prosper Insights & Analytics survey, across all generations, 27% of respondents indicated that they were already using AI. Further in a business context, research from Deloitte Consulting LLP's 2024 'The State of Gen AI in the Enterprise report: Now decides next' indicates that 78% of organizations surveyed expect to increase AI spending in the next fiscal year.
Prosper - Heard of Generative AI
For many enterprises, one critical use of AI will be to augment incident response management. Managing the response to critical IT incidents is about detection, speed, and accuracy. It's also about context, communication, and strategy. AI-driven tools can help with the first part. However, it takes keeping humans—and their empathy—in the loop and in the lead to handle the second part and effectively manage the incident response process.
AI can accelerate and improve incident detection, but context matters
AI-driven tools are revolutionizing the process of incident detection with their ability to rapidly discover system anomalies and other technical issues faster than humans can—in real time in most cases. They can holistically assess system behavior—at scale—and they can also prioritize incidents by criticality. They can even recommend automated remediation steps. Human teams can't perform at this level.
However, as good as AI is at spotting incidents, it can't put those incidents in a business context; it can't tell you why they matter. 'For example, AI can flag an API latency spike in real time,' says Faruk Muratovic, principal, Deloitte Consulting LLP and US AI & Engineering Strategy & Services Leader. 'But it can't understand the context of that spike. It can't tell that the unusual spike is happening during a critical, time-sensitive interaction with your highest-value client.'
In moments like these, human judgment is crucial. AI might understand the technical issue, but only humans can understand the wider context of the issue—whether that issue has regulatory implications, involves a high-value client, or is interrupting a mission-critical workflow. Without humans in the loop, to apply a value judgment to issues, organizations can fall into the trap of underreacting to major incidents or overreacting to minor ones. The upshot? Contextual appreciation isn't an AI feature. It's a core human domain.
Automation should be coupled with human communication
In addition to contextual awareness, humans play a critical role in communication 'awareness.' In other words, humans don't do the work of automation, but they control how the messaging around how that work is delivered and perceived.
For example, in critical incident management, AI tools can perform an array of tasks, such as traffic control, services management, patch management, or systems isolation that help response teams resolve the incident. Yet even though a solution might technically be correct, things can go sideways quickly if the communication about the incident, and its solution, is delivered without context or nuanced awareness of the audience.
'With incident management, it's not simply about finding and resolving an issue—it's about what and how you communicate about the issue,' Muratovic stresses. 'Stakeholders from clients, to the C-suite, to regulators need clear, honest assurance that humans are aware and in control of the situation.'
Only humans can put the incident in perspective to anticipate what's next based on the business context, regulatory environment, and reputational considerations. Only they can ensure that the message is unified across the organization so that everyone is speaking with the same voice.
Incident management strategy is a human endeavor
With AI, incident triage is undoubtedly faster, so response teams save valuable time in formulating and implementing a solution strategy. But that strategy should be human led. In complex or regulated environments, speed matters, but so does cross-functional orchestration that accounts for the nuances of the situation. According to Deloitte's Muratovic, humans take incident response from a tactical firefighting exercise to strategic resolution.
'AI might tell you where the fire is,' says Muratovic, 'but it takes a human in the loop and in the room to understand and make the call on whether you need a fire extinguisher or an evacuation plan.'
For example, the incident might be a disclosure issue for a compliance officer, whereas a product manager may recognize that the incident could affect an ongoing, sensitive client negotiation. Obviously, DevOps, Security, PR and other functions might be involved, depending on the incident. The point is that only humans can orchestrate and contextualize the response and align it with business strategy and continuity goals.
Humans will lead the future of incident response
As AI's capabilities mushroom, organizations will be tempted to view it as an incident-response panacea. The reality is, however, that for all AI's power, it lacks human contextual awareness, empathy, and strategic intuition. AI might excel at handling intricate tasks like pattern recognition, automated alerting, and even executing more complex, predefined processes. But the critical, intuitive concepts like what matters, why it matters, how to prioritize, when to escalate, and how to position for recovery will remain in human hands—perhaps indefinitely.
'The goal shouldn't be to replace humans with AI,' Muratovic clarifies. 'It should be about giving them tools to augment their work and keeping them where they add irreplaceable value: decision-making, communication, and strategic leadership in the crucible that is incident management.'
That augmentation is a powerful partnership. With AI detecting, exposing, and routing incidents, humans can focus on developing, prioritizing, orchestrating, and communicating the response.
The takeaway, and the future
Successful incident response management requires more than technical help; it's a partnership that pairs human intuition, insight, empathy, and communication with AI's power to automate the process. As AI capabilities and tools continue to improve, the human effort required for context interpretation, oversight and intervention will likely decrease. That doesn't mean that humans won't be in the loop, just that the AI/human partnership will continue to improve incident management in ways that neither could accomplish alone.