AI for Hong Kong AIs: Begin With a Governance Framework

Share this post:

Artificial intelligence (also known as Augmented Intelligence) is becoming more than another popular buzzword. The emerging technology is becoming part of our daily lives, with many enterprises viewing it as the new driver for business growth.

As enterprises speed up their adoption of AI in business, the spotlight shifts to its governance. In banks, decision makers and senior management team members are calling for a better governance framework along the lines of FAT (fairness, accountability and transparency).

Regulators are taking note. Hong Kong’s currency board and de facto central bank, the Hong Kong Monetary Authority (HKMA), published its guidance to Hong Kong Authorized Institutions for customer protection when using big data and AI (BDAI) along with high-level principles on AI. Authorized Institutions include banks, restricted license banks and deposit-taking companies.

The guidance advises Authorized Institutions to be mindful in planning, designing and implementing a wide range of BDAI applications that use personal data from customer interactions. The data can help BDAI applications to improve business efficiency, simplify customer risk profiling, pinpoint money laundering activities and detect fraud.

This guidance answers three important issues that face all BDAI initiatives in banks:

  1. Enforcing a code of practice covering customer data protection, corporate accountability, fairness and explainability of outcomes, and actions from BDAI.
  2. Determining the method for certifying BDAI solutions that meet the guidance requirements.
  3. Identifying products or services for testing how well the BDAI solutions comply with the high-level principles on AI.

BDAI applications in banking do not operate alone and often involve many diverse groups working together. These can range from business leaders and frontline staff to IT vendors and in-house IT teams. An established guidance can help to set the rules before adverse customer sentiment develops from bad experiences with early adopters.

A cornerstone framework can help Authorized Institutions to achieve AI at scale. IBM’s AI@Scale offers one that comprises 6 dimensions: Strategy, Operating Model, Data & Platform, Operations, Change Management and People & Enablement.

The framework also offers a rich set of solution assets, including a maturity assessment model, a reference architecture and IBM’s best-in-class AI governance tools, such as IBM OpenScale and AI Fairness 360 Toolkit. Together, they help Authorized Institutions to take a holistic route to addressing HKMA’s 12 guiding principles.

Authorized Institutions can use a four-step approach to deploy the AI@Scale framework:

  • Step 1: Assess the gaps in the existing process, methods and tools
  • Step 2: Evaluate and build an enterprise-wide framework
  • Step 3: Adopt the framework with quick-win pilots
  • Step 4: Scale to implement the AI ecosystem in production

The age of AI is already here. Far from another industry buzzword, AI is already offering a multitude of benefits for Authorized Institutions that remove inefficiencies, take user experience to the next level and offer new revenue opportunities across the banking value chain.

Proper AI governance can help to navigate the many challenges and issues concerning bias, personal data usage, explainability and compliance. It is a must-have if you do not want your AI journey derailed.

IBM Distinguished Engineer and Chief Technology Officer, Hong Kong

Joseph Ma

Associate Partner, Cognitive & Data Analytics Leader, Global Business Services, IBM Hong Kong

More AI stories

機密運算杜絕非法存取數據 助建立可信企業雲端營運環境

(文章於2021年7月2日在香港經濟日報網站刊登) 數據安全,是金融和電訊等需要處理敏感資料的行業,採用公共雲及混合雲服務的一個主要障礙。一般雲端服務商可以為處於靜止儲存狀態的企業數據,進行高強度加密,盡量減低被破解的機率;但是在運算和處理前的一瞬間,這些數據還是必須被解密。隨著科技日新月異,這種以往看來安全的狀態越來越可能被不法分子有機可乘。

Continue reading


(文章於2020年6月29日在香港經濟日報網站刊登) 金融服務業一向被視為應用科技上相對保守的行業,客戶利益、監管法規、保安風險等因素都促使行內企業要審慎行事。但近年來消費者渴求切合需求的金融服務,而且顛覆式的金融科技也正向傳統金融服務步步進逼,銀行、保險公司、證券行等都要積極回應挑戰。

Continue reading

2020 IBM X-Force 威脅情報指數報告詳解網絡攻擊最主要的三大初始媒介

IBM Security 於上週公佈了 2020 年 IBM X-Force 威脅情報指數報告。報告重點闡述了數十年來犯罪技術經歷了怎樣的演變,在此期間網絡犯罪技術非法訪問了數百億條企業記錄和個人記錄,並利用了數十萬個軟件缺陷。報告顯示,在首次遭受攻擊的受害者中,有 60% 是源於過往被盜憑證或已知軟件漏洞,攻擊者無需大費周章實施詐騙就能獲得訪問權限。

Continue reading