Hong Kong is lagging in the AI race – can it raise its game, and how?

In the third installment of the series previewing Hong Kong's five-year plan and policy address, set for announcement on September 16, reporters Leopold Chen and Theodora Yu examine the city's potential to enhance its adoption of artificial intelligence (AI). The piece revisits earlier reports, including those published in parts one and two.

One individual's encounter with Hong Kong's inefficient service delivery illustrates the challenges the city faces. Former lawmaker and resident Michael Tien Puk-sun reported that the government's 1823 hotline, intended to address public complaints, failed to log his initial report regarding illegally dumped construction waste. Unwilling to accept the government's inaction, Tien publicized his frustration on social media.

In response to residents' grievances, the government has pledged to integrate AI into public services, including the 1823 hotline. This integration is slated to be a focal point when Hong Kong unveils its first five-year plan and latest policy address on Wednesday. While experts acknowledge significant hurdles, they urge the authorities to bolster their risk tolerance, refine data governance frameworks, and ensure local tools remain compatible with international standards.

Last July, Hong Kong introduced three AI-driven tools – HKChat, HKPilot, and HKMeeting – on its private cloud service to elevate civil service efficiency. HKChat facilitates smart searches for government workers, HKPilot assists in document drafting, and HKMeeting transcribes meetings. However, the city's progress lags behind neighboring regions.

Singapore introduced its Pair smart search chatbot for public officers in 2023, which has garnered over 80% adoption among civil servants and includes more than 7,700 agency-specific bots. Additionally, the platform offers SmartCompose for drafting public inquiry responses and Transcribe for speech-to-text recording, featuring a comprehensive public inquiry AI chatbot accessible to citizens.

Singapore's commitment to AI development was solidified with the establishment of the National AI Council in February, chaired by Prime Minister Lawrence Wong, to outline the country's AI strategy. Hong Kong also possesses a comparable Committee on AI+ and Industry Development Strategy, chaired by Financial Secretary Paul Chan Mo-po, which convened for the first time in late June.

Hong Kong's Digital Policy Office, under the Innovation, Technology and Industry Bureau, is spearheading AI-driven transformations across government departments. The office plans to roll out a public-facing chatbot on the iAM Smart digital identity portal later this year and intends to integrate AI into the 1823 hotline, enabling transcription of calls and automatic case summaries to free up frontline staff for more complex cases. This initiative is part of a broader lineup of 30 AI-enhanced public service improvements.

Duncan Chiu, a lawmaker representing the innovation and technology sector, emphasized that the government's adoption of AI across industries and its role in driving future development is a top priority. Nonetheless, Chiu highlighted that a historically privacy-focused legal framework has impeded the sharing of resident data between different government units.

He advocated for a risk-based, tiered classification mechanism for AI applications within the civil service over the next five years, encouraging officials to be more proactive in implementing new technologies in lower-risk scenarios while maintaining caution in high-risk areas.

Chiu also called for the refinement of a legal framework around data governance, stating that once such a framework is in place, it would clarify the path forward for AI development. Professor Ray Cheung Chak-chung, an associate provost in digital learning at City University of Hong Kong, noted that while Hong Kong could benefit from a "latecomer's advantage," it confronts critical challenges in digital literacy and risk tolerance.

He recommended the government adopt a clear "trial-and-error" policy to foster AI experimentation, stressing that transparent communication with stakeholders would mitigate potential mistakes and maintain course even in the face of less-than-ideal outcomes. Cheung argued that highlighting immediate benefits of AI-powered public services could bolster societal support for AI adoption.