Hong Kong finds Gen‑AI value gap

Generative AI is in use across Hong Kong — especially inside banks, insurers, and asset managers. That’s the concrete part. The catch is the payoff — firms report lots of pilots but only patchy, measurable business results. The Hong Kong Institute for Monetary and Financial Research put the gap on paper in April 2025 — a clear snapshot of adoption without scaled value. (hkma.gov.hk) Why does the report matter now? Because Hong Kong is trying to industrialize GenAI safely — rules, funding, and a local foundation model effort are already in motion. If companies can’t turn experiments into outcomes, the city’s push to capture economic gains stalls. (digitalpolicy.gov.hk) What did the HKIMR study actually find? Three-quarters of the 55 financial firms surveyed had either rolled out a GenAI use case, were piloting one, or were actively exploring investment. Most of that activity is internal — employee assistants and back-office efficiency — not enterprise-level revenue lifts. (hkdca.com) Is this a local quirk or a global pattern? It’s a pattern. The problem shows up everywhere — lots of tool adoption, far fewer examples of redesigned workflows or measurable EBIT gains. Companies can build flashy pilots quickly — scaling them so they actually move the needle is harder. (mckinsey.com) Why aren’t pilots turning into value? Three practical bottlenecks. Leadership time and accountability are thin — projects sit with IT or pockets of power users. Data and governance are fragmented — firms lack clear guardrails. And the skills shortage means projects stall before process changes stick. Those constraints show up in both the HK reports and broader industry surveys. (hkdca.com) What are Hong Kong firms using GenAI for today? Mostly internal tasks — virtual assistants for employees, document summarization, and automation of routine workflows. Customer‑facing deployments exist, but they are fewer and more cautious because of accuracy and compliance worries. That internal-first pattern explains why topline impact is still limited. (fintechnews.hk) What are regulators and the government doing about it? The Digital Policy Office published technical and application guidance to steer safe development, and the city has funded local GenAI R&D to build capacity and standards. The goal is to nudge firms from pilots to governed, scalable deployments. (digitalpolicy.gov.hk) What should companies do next? Stop treating GenAI as a toolkit and start treating it as a process change. Put a senior sponsor on ROI metrics. Redesign workflows where AI sits — not bolt tools onto old processes. Invest in data plumbing and governance so models are reliable and auditable. Train and reward people for changed ways of working. These are boring steps — but they’re what turn experiments into value. Bottom line. Hong Kong has the tech appetite and the policy scaffolding — the missing piece is disciplined execution: leaders, metrics, and governance that force pilots to pay rent. (hkma.gov.hk)