In recent years, the widespread adoption of artificial intelligence (AI) around the world has brought benefits to society on different aspects, making it well recognised as an irreversible trend to enable development and transformation. As outlined in the 14th Five-Year Plan, the Central Government indicates clear support for developing Hong Kong into an international innovation and technology (I&T) hub, and better integrating the city into the overall development with the Mainland with the Guangdong-Hong Kong-Macao Greater Bay Area (GBA) as an entry point. Riding on the opportunities brought about by the rapid development of e-commerce and the AI trend across the globe, the Hong Kong SAR Government (“the Government”) is also proactively promoting the development of Hong Kong into a smart city fuelled by digital transformation.
Although currently there is no universal definition of AI, it is fundamentally a series of computer programmes and machine learning that mimics the problem-solving and decision-making capabilities of human beings. The rapid global development of AI can be seen as a double-edged sword. On the one hand, it brings economic benefits, improves the quality of life, and handles high-risk tasks. On the other, it could pose risks to the public if used inappropriately. In the process of developing AI, Hong Kong has likewise encountered both the ensuing opportunities and challenges. Key stakeholders, such as consumers and traders, generally have an insufficient understanding and experience in AI technology, which impact their confidence and attitude when using AI. With a view to encouraging stakeholders in Hong Kong to embrace the sea change towards I&T with a positive and proactive attitude, thus achieve a sustaining and win-win situation, the Consumer Council (the Council) conducted its first-ever study on the use of AI in e-commerce in Hong Kong, titled “Fostering Consumer Trust – Ethical AI in Hong Kong” (the Study). Adopting a forward-thinking approach, the Study aims to identify the various challenges currently faced by Hong Kong’s consumers and traders when they use or develop AI, as well as their opinions; to review the objectives and policies of AI governance adopted by different countries and regions around the world; and to listen to the views of stakeholders. Through the findings of the Study, the Council advocates a series of recommendations to enable the healthy development of AI in e-commerce, on the basis of complying with ethical standards and balancing the interests of all parties, including safeguarding consumer rights.
The Council stresses that in order for AI to thrive in Hong Kong, apart from strengthening consumer education on issues pertaining to e-commerce, including data privacy and cybersecurity risks, traders should take a proactive step to adopt industry best practices in formulating company AI policy and corporate governance, as well as establish a “Consumer Charter” to safeguard consumer rights. Furthermore, the Government could establish a holistic policy for long-term AI development, so as to progressively drive digital transformation in society. The Council believes that through a three-pronged approach – policy and regulation, corporate governance and consumer education – Hong Kong can be developed into a world-class smart city.
Recognising the Pros and Cons of AI
Enhance Consumer Confidence Through Effective Risk Mitigation
When applying AI in e-commerce, traders use consumer data with AI algorithms at the back-end operations for their own benefits, which might result in untransparent but significant implications for consumers in terms of product pricing, choices, and search results. There have been controversial cases in other jurisdictions. For instance, traders had adopted personalised pricing with AI targeting consumers of a certain age range or characteristics, which could turn into a practice of price discrimination. Besides, higher search rankings generated by biased AI algorithms could manipulate consumers’ choices as well. These might distort market competition and raise consumers’ scepticism towards AI. The Council is highly concerned about the imbalance between consumers and traders in terms of information possession and bargaining power, a phenomenon which has become increasingly prevalent amidst the rapid development of AI, and thus opines that such ethical issues must be nipped in the bud and promptly addressed.
The scope of the Study focuses on business-to-consumer (B2C) e-commerce and adopts a mixed-method approach comprising a quantitative online consumer survey; review of popular e-commerce platforms (including AI applications and public information on websites) amongst local consumers; in-depth interviews with 19 major e-commerce traders, technology providers and industry experts in Hong Kong; desk research on relevant policy guidelines and initiatives on the use of AI in 10 selected jurisdictions; as well as review of consumer complaint cases. The Study aims to provide insightful recommendations to the Government, traders and consumers to promote the responsible and ethical use of AI in the e-commerce industry.
Keen on Online Shopping but Unfamiliar with AI
Consumers Anticipate Right to Know and Right to Choose
To comprehend consumers’ perception and usage habits of AI and data privacy, the Council conducted an online quantitative consumer survey with 1,219 respondents in Hong Kong aged 15 or above who have visited local or non-local online stores. The survey revealed that despite local consumers being quite active in e-commerce (77% of the respondents visited or made purchases at online stores on a daily or weekly basis), their familiarity with AI is still low (up to 75% of the respondents expressed that they were unfamiliar with AI). Besides, consumer satisfaction and trust towards the overall adoption of AI tools and individual applications varied vastly. For example, only 31% of the respondents trusted AI, 41% agreed that AI addressed their needs accurately, whereas slightly more than half (57%) of the respondents agreed that AI helped reduce their time spent for choosing products.
Moreover, the survey reflected consumers’ concerns and worries towards various issues such as personal data privacy, their right to know and right to choose, and the accountability of AI. 74% of the respondents said that they were worried about the excessive data collection of traders, while 72% were worried that if the AI algorithm goes wrong, no one will be responsible for the problems caused. Consumers also anticipated more information transparency and the choices to opt-in and opt-out. 78% hoped to be informed by traders when they are using AI to provide services, while 81% urged to have the right to opt for the use of AI tools.
Notwithstanding consumers’ concerns about cybersecurity and privacy protection, they do not act cautiously in practice when shopping online. For instance, up to 60% of the respondents had never read privacy policies at online stores or simply did not know what privacy policies were, while the reasons for not reading the privacy policies included the content being too lengthy and wordy, being too hard to understand, or that the consumer believes that the online stores can gather their personal information online regardless of whether they read the privacy policies or not. Amongst those who had read the privacy policies before, 43% of them had stopped browsing an online store, mainly because too much of their information would be gathered, too many third-party organisations would receive their personal data and the number of purposes for which their data was collected was excessive. However, should consumers stop browsing a website after assessing the risks involved, it is also an effective way to protect themselves. Furthermore, when online stores display a pop-up message to obtain consent from consumers to collect their cookie information, 27% of the respondents would simply neglect these messages or just leave it there. The Council emphasises that to foster more cautious online shopping habits and thus to ensure consumers put their knowledge into practice and take cybersecurity measures to protect themselves, consumer education is indispensable. 77% of the respondents agreed that public education on AI should be strengthened and more information on AI should be provided.
Accuracy and Stability are Key for Enhancing the Satisfaction of AI Tools
The Study also examined consumers’ usage and satisfaction with 4 types of AI tools. In terms of usage, product recommendation had the highest usage rate (89%), followed by chatbots (75%), advanced biometrics (55%) and augmented reality (AR) (32%). Yet, when it came to consumer satisfaction, the 2 more popular AI tools had lower ratings, with only around half (56%) of the respondents happy with product recommendation, whereas chatbots had the lowest satisfaction with only 38% respondents deeming it useful for improving their online shopping experience, believed to be due to the chatbots’ inability to accurately address consumers’ questions. Moreover, close to half (48%) of the respondents were particularly cautious about providing personal information when using chatbots, around 10% higher than the other 3 AI tools. In fact, it has become increasingly prevalent in recent years for traders to provide customer services with chatbots, as not only could it help them cut back on labour costs, but it could also provide 24/7 customer support. However, inaccurate answers from chatbots not only fail to solve the problem, but also largely discount the satisfaction of consumers. On the contrary, over 70% of the respondents were more satisfied with AR (74%) and advanced biometrics (73%) respectively, mainly attributed to the more stable performance, and more gratifying and authentic experience offered.
Privacy Policies of Online Stores Show Much Room for Improvement
The Study also reviewed the privacy policies of 112 local and non-local e-commerce platforms. Although 90% of the reviewed online stores disclosed their ways, purposes of data collection, and types of data collected on their websites, there were various shortcomings. For example, while AI is widely adopted, only 6 reviewed traders mentioned the training of AI or machine learning as their purpose of data collection. Moreover, 6% to 10% of reviewed online stores would also collect data that is not directly relevant to the transaction, such as the consumer’s education, income, marital status and employment status, etc., which might be considered excessive. Traders should also enhance transparency when handling consumers’ data. For instance, only 41% mentioned they would anonymise the data before using it for data analytics; only 17% were specific about the data retention period. The reviewed online stores also lacked clarity in stating the rights of consumers, as merely 42% informed consumers how to reject personal data collection or opt out of cookie tracking. The findings reflect the need for online stores to evaluate and improve their protection of consumers’ data privacy.
Anticipating Increased Support to Tackle Challenges in Talents, Funding and Data
Based on the 19 in-depth interviews conducted by the Council with the management of e-commerce traders, technology providers and industry experts, the general opinion was that local traders still have insufficient understanding of AI, and did not fully grasp its strategic advantages. Additionally, traders encountered various challenges when developing AI in Hong Kong, making it difficult for AI to flourish on a broader scale.
The interviewees opined that the adoption of AI in Hong Kong is still at an early development stage. They graded the current AI development in Hong Kong 4.9 out of 10 on average, where a score of 10 refers to extremely advanced development in the region. The interviewees expressed that currently, traders in Hong Kong mainly adopt AI to assist business operations, with the main reasons being, in the order of priority, to provide better customer services, improve efficiency, and reduce costs. However, they still do not view AI as a strategic differentiator to generate new income. On the other hand, traders encountered 3 key challenges while adopting AI, namely the scarcity of talents, insufficient funding, and limited data. As the e-commerce market in Hong Kong is relatively small, the data and its variety available for AI model training are probably insufficient, posing big challenges for Hong Kong businesses, in particular small and medium-sized enterprises (SMEs). This might affect their long-term competitiveness in the market.
The majority of stakeholders and interviewed traders reflected an insufficient understanding of AI amongst the industry, and expressed their hopes that the Government could dedicate more resources in the application of such technologies in the future, so as to assist enterprises in developing the application of AI. At the same time, the industry could also encourage traders to open up their own data and share with one another to facilitate the growth of AI. To address the insufficiency of data, industry players can make use of other unstructured data, such as speech and image data, to train their AI models, or consider forming partnership with other traders or universities, locally or internationally, so as to expand the database via multiple channels to tackle the challenges in growing AI innovation in Hong Kong.
Reference the Experience of Other Jurisdictions to Expedite a Local AI Governance Framework
To effectively develop AI and mitigate relevant risks, currently over 160 sets of AI ethical principles or guidelines have been introduced by various international organisations, while a number of jurisdictions have developed, or are in the process of developing, their own national AI regulations. The Council reviewed the AI development blueprints and governance frameworks of 10 selected jurisdictions, namely the Mainland, Canada, France, Germany, Japan, Singapore, South Korea, the European Union (EU), the United Kingdom (UK) and the United States (US). It was found that various common themes relating to consumer rights were regulated and monitored, including accountability, transparency and interpretability, data privacy, etc. For example, the Mainland’s “Administrative Provisions on Algorithm Recommendation of Internet Information Services”, effective from March 2022, seeks to regulate algorithms, and monitor traders employing algorithms for recommendation purposes in their online stores, search filters, or social media. On the other hand, the EU adopts a risk-based approach in its proposed AI Act in April 2021, which bans specific uses of AI with “unacceptable risks”, and heavily regulates some other uses that carry “high risks”. Furthermore, both the Mainland and the EU have incorporated the regulation of automated decision-making of data users in their data privacy laws for better consumer protection, allowing consumers to have the right not to be subject to a decision based solely on automated processing. Japan and South Korea conducted regular surveys on the status of consumers’ understanding of AI; Singapore focused on public education to increase AI literacy, and also provided various support to traders, such as launching an “A.I. Verify” toolkit for traders to verify their AI performance, so as to promote the practice of ethical principles. As AI will become an important enabler of digital and economic development, the Council recommends the Government to make reference to these governing experiences in other jurisdictions to accelerate the development of Hong Kong’s own governing framework on AI.
To advocate the collective development of “responsible and ethical AI” by all stakeholders, in an effort to safeguard consumer rights and foster long-term societal and economic benefits, the Council puts forward the following recommendations:
Recommendations for Traders:
Enhance Corporate Governance to Safeguard Consumer Privacy
Establish “Consumer Charter” to Increase Public Confidence
E-commerce traders should ride on the opportunities from the thriving development of AI to strengthen corporate governance, which could in turn foster consumer trust. When employing AI, traders should adopt the following 8 good trade practices:
- Formulate a clear company policy on AI, data usage and storage with good compliance and disclosure;
- Dedicate resources for AI development with clear accountability, e.g. appointing a person to be accountable for the ethical use of AI;
- Validate AI models before deployment;
- Ensure consumers’ data is handled in a safe and secure manner to protect consumers’ data privacy;
- Communicate with staff and ensure they follow the company policy and ethical standards of AI;
- Foster communication with consumers;
- Provide choices for consumers to choose whether to use AI; and
- Clarify clear responsibility between traders and third-party technology providers, and ensure data is handled and transformed safely and ethically.
As many SMEs might have limited resources to establish their own set of AI policies, the Council suggests industry associations develop a “Consumer Charter” and invite their member traders to commit to and follow. Apart from providing guidance on data governance, the “Consumer Charter” will also cover the rights of consumers, system security, opt-in and opt-out options, as well as reporting of data breaches, dispute resolution mechanism, and dedicated and clear disclosure of AI policy. By complying with the “Consumer Charter”, traders can strengthen their trade practices and gain consumer confidence. The Council recommends the following 7 key points for the “Consumer Charter”:
- Do not manipulate consumers’ choices;
- Be fair to all consumers and do not discriminate against consumers;
- Provide choices for consumers to opt in and opt out of the use of AI easily with immediate effect;
- Ensure consumer requests can be addressed;
- Establish an effective dispute resolution mechanism in case of non-compliance; and
- Collect data at an adequate level.
Recommendations for Consumers:
Cultivate Good Online Shopping Habits Strengthen Self-protection Awareness
This Study has revealed the significant need for consumers to strengthen their self-protection capacity when shopping online, especially against the risks of the abusive use of personal data. The Council recommends the following 7 practical and simple tips for consumers to shop smartly online:
- Choose the right e-commerce platform and make it a habit to always read data privacy policies, and terms and conditions;
- Pay attention to website updates, and thoroughly read the pop-up notices, news and updates when entering a platform;
- Make the best decision against tracking of locations, browsing histories, etc;
- Provide adequate personal information only;
- Keep a good record of membership accounts;
- Check privacy settings regularly;
- Exercise your rights against automated decisions when needed.
Recommendations for the Government:
Build a Fair and Competitive E-commerce Market
Establish Development Policy and Nurture Public Understanding
Both consumers and traders have expressed their wish to obtain more knowledge relating to AI. The Council recommends that from a societal perspective, education campaigns could be launched to strengthen consumers’ understanding of AI in various facets, including its basic concept and operations, potential benefits and risks, consumer rights regarding the use of AI, related guidelines consumers can refer to, and best practices in the marketplace, etc. This Study classified consumers into 4 segments based on their attitude towards AI, namely “supportive consumers”, “prudent consumers”, “unready consumers” and “unfavourable consumers”. The Government and various stakeholders could consider targeted education strategies according to each of these 4 segments’ characteristics, so as to establish a correct concept in the use of AI, as well as to nurture an AI culture and trust in society.
The Government has outlined the development of AI in the “The Smart City Blueprint for Hong Kong 2.0”. Currently, AI-related initiatives and guidelines are championed by respective authorities based on their own assessment on industry adoption and governing practices. To further empower different industries to practise the ethical use of AI, relevant regulatory bodies and Government bureaux/departments could establish sector-specific guidelines or standards for AI adoption for traders to follow. Furthermore, traders’ AI compliance results could be disclosed for public surveillance, while consumer feedback and suggestions for improvement of such initiatives could be collected.
At the same time, the Study findings show that consumers, traders and stakeholders all recognise the need for the Government to establish a holistic and long-term policy framework, with a clear development plan for AI. Making reference to relevant international regulatory measures and experiences, the Government could consider covering the 7 areas below when formulating the AI plan:
- Establish clear vision and key milestones for AI development;
- Lead by example to accelerate digital transformation;
- Increase funding to support AI projects to commercialise;
- Enrich open data and encourage utilisation of data;
- Assist traders to build business connections with partners in the Mainland and in the globe;
- Attract and nurture AI talents; and
- Provide AI ethics training or guidelines to traders.
In the digital economy, companies with more data are obviously of a better competitive advantage. To reduce disparities between SMEs and big techs, the Council recommends the Government and the industry facilitate the sharing and utilising of data among traders, thus helping SMEs develop and train their AI models. Meanwhile, to ensure a fair and competitive market in the e-commerce industry and to safeguard consumer interests, the Government must take appropriate measures and prevent traders from using big data to manipulate the market for their own advantages.
Concerted Effort to Realise the Vision of a Smart City
Amidst the rapid transformation of the digital economy, AI has become a driving force for societal and economic development. Different countries and regions have formulated AI development roadmaps covering various facets, such as governance, digital policies, industry support, consumer safeguard and education, with the aim of fully unleashing the potential of AI and effectively mitigating its risks. Strategic directions around the globe generally always include “responsible and ethical AI” policies to build stakeholders’ confidence and positive attitude in using relevant technology. The Council hopes that while the Government is actively promoting digital transformation and AI development, it can make reference to the experiences and approaches of other jurisdictions and establish an AI development plan and regulating policies to encourage the development of “responsible and ethical AI” in society and ensure consumer interests are well protected. The Council will actively participate and work with various stakeholders. The Council believes that a three-pronged approach must be adopted, including continuously improving policy framework, improving the performance of digital governance amongst traders, and strengthening consumers’ ability to protect themselves, in order to leverage and transform the power of AI into an engine to fuel the development of Hong Kong.
Visit www.consumer.org.hk/en/advocacy/study-report/ai_in_ecommerce to view the full electronic version of the “Fostering Consumer Trust – Ethical Artificial Intelligence in E-commerce” report.