How big businesses are handling the roll out of Generative AI (www.techradar.com)

Generative AI’s enterprise promise is colliding with reality: after nearly three years of hype, large organizations are sliding into a “Trough of Disillusionment” as pilots struggle to show measurable ROI, and Gartner warns up to 30% of GenAI projects may be abandoned by 2025. Common technical and operational blockers include poor data quality and provenance, insufficient risk controls, model hallucinations and bias, spiraling costs, talent gaps and immature AI literacy — all compounded by emerging regulatory standards like the EU AI Act. These issues show GenAI adoption is as much a people-and-process challenge as a technology one. Successful scaling requires a disciplined, product-aligned operating model and a three-phase path: Discovery & Baselining (assess data landscape, tech stack, maturity, and define success metrics), Tooling & Design (select models and infrastructure—cloud or on-prem—architect for security/governance, integrate with workflows and UX), and ROI & Scaling (pilot, measure against KPIs, then expand). Responsible AI must be embedded throughout: data provenance checks, bias and safety testing, policy enforcement, access controls, human-in-the-loop for high-risk decisions, continuous monitoring, audit trails and model re-evaluation. Real-world wins — faster software testing in banking and a pharma document-review assistant achieving >95% accuracy and 65% lower manual effort — show value is achievable, but only by aligning tech, governance and change management across the enterprise.