Why AI is hype - from an IT Operations POV. And why companies will hire back engineers

A veteran IT operations director leading teams at major global tech companies has published a viral critique arguing that current AI, particularly 'agentic' systems, is overhyped and fundamentally unsuitable for real enterprise automation. Despite initial enthusiasm and significant corporate investment, his extensive testing reveals AI fails beyond simple, well-defined tasks like enhanced search or first-tier support. The core failure point is the uncontrollable hallucination problem inherent to LLMs, which lack true logical reasoning or error-correction capabilities. This makes them unreliable for the complex, risk-averse workflows that define corporate IT operations, where the primary job is systematically removing errors, not introducing new ones.

He contends that the demonstrated 'reasoning' of AI agents is merely 'iterative predicate generation'—a sophisticated pattern-matching trick, not genuine logic. While AI can automate repetitive, cookie-cutter processes, it collapses when tasks require actual thinking or adaptation. From a corporate governance perspective, this is a critical flaw. Companies invest heavily in frameworks like ISO standards and Change Management to minimize risk, yet AI agents reintroduce error at an unacceptable scale. The author predicts this realization will trigger a shift, with companies scaling back AI automation projects and rehiring human engineers to handle the nuanced judgment and risk mitigation that current systems cannot replicate.