China Business Knowledge at CUHK

14/09/2026

Companies are increasingly seeding synthetic content online to influence generative engine optimisation (GEO). While synthetic content can boost visibility, it also risks consumer scepticism, since it is often perceived as manipulative.

“AI engines work like a ‘black box’, but some consumers have an expectation that content manipulation exists on the internet, so they do not always blindly follow what GenAI recommends,” says Liao Chenxi, Associate Professor of Marketing at CUHK Business School.

In collaboration with Tony Ke, Professor of the Department of Marketing at the School, Professor Liao finds that firms with high-quality products are less inclined to deploy synthetic content. Although lower-quality firms consistently benefit from such tactics, simply banning such content may not always improve consumer welfare.

Read more: https://cbk.bschool.cuhk.edu.hk/can-you-truly-trust-ai-recommendations

The digital world is awash with AI-spun content as firms exploit AI engines for visibility, but Professors Liao Chenxi and Tony Ke find this game could backfire

05/09/2026

As corporations increasingly entrust high-stakes resource allocation and inventory management decisions to AI, relying solely on advanced models can inadvertently introduce operational risks. These powerful tools may inherit, and even amplify, human biases.

A new study co-authored by Professor Chen Zhi, Associate Professor in the Department of Decisions, Operations and Technology at CUHK Business School, explores the “paradox of intelligence”, proving that superior brainpower does not guarantee superior commercial choices.

📌 AI models consistently mirror human ordering biases, ordering too little in high-margin settings and too much in low-margin ones.
📌 The most advanced model over-ordered by 70% in low-margin scenarios, talking itself out of optimal quantities via elaborate over-analysis.
📌 Failure modes vary across architectures, spanning overthinking, sticking too strictly to simple rules, and limited computing power.
📌 Managers must deploy explicit and clear rule-based prompting, choose the right AI model for specific tasks, and ensure a human-in-the-loop to constrain biases.

Read the full article: https://bit.ly/4qBVTeo

Follow us for more

03/09/2026

AI agents can act as humans in experiments and social simulations, but can they really think like us? While large language models (LLMs), the systems behind AI chatbots, are getting more advanced, they still sometimes act like robots and generic prompts often just don’t cut it.

“LLMs in behavioural and social simulations can fail not only because of flawed reasoning, but also because of the limitations in how they interpret their tasks. Even if they understand the tasks, their reasoning process may not align with human thinking,” says Philip Renyu Zhang, Professor in the Department of Decisions, Operations and Technology at CUHK Business School.

In a new study co-authored with Jenny Jin, Assistant Professor in the same department, Professor Zhang finds that if an AI’s initial understanding of the social rules, goals, and context is incomplete or inaccurate, even its best reasoning will lead to decisions that don’t match human behaviour.

Therefore, they introduce a two-stage approach that explicitly guides AI to build an accurate mental map of the task and then reason within it, a critical step for reliable AI simulations, especially in intricate human interactions.

Read more: https://cbk.bschool.cuhk.edu.hk/can-ai-really-mimic-our-decision-making/

LLMs can successfully reproduce simple social simulations, but often fail to act like human in complex scenarios, Professors Philip Zhang and Jenny Jin find why

31/08/2026

The rise of AI coding assistants signals a new era in software creation, but can these tools truly fast-track developer work and careers?

Professors Keongtae Kim and Michael Zhang from the Department of Decisions, Operations and Technology at CUHK Business School investigate how AI coding assistants like GitHub Copilot X reshape software developer work patterns and reveal some fascinating shifts.

AI coding assistants significantly boost developer activity and encourage exploration of new languages, especially for casual coders. Using AI tools can actually lead to faster internal promotions. However, this speed comes with a trade-off: more copy-pasted code and less originality in projects for less established users.

So, is AI democratising code, or are we trading quality for quantity? Read more: https://cbk.bschool.cuhk.edu.hk/is-ai-creating-more-productive-but-dull-programmers/

28/08/2026

Overtourism and visitor misconduct strain destinations worldwide, prompting site managers to search for improved visitor management methods. A new study co-authored by Professor Robert Li Xiang, Fung King Hey Memorial Professor of Tourism Management and Director of the School of Hotel and Tourism Management at CUHK Business School, indicates that intelligent robots can serve as highly effective automated persuasion agents to foster responsible conduct without creating psychological tension.

📌 By providing uniform and non-confrontational reminders, smart robots enable policy makers to effectively address the complexities of visitor governance.
📌 Activates visitor self-awareness through impression management theory by creating a noticeable yet non-confrontational presence.
📌 Simpler-looking robots work best with a direct message, while human-like models succeed using humour.

Read the full article: https://bit.ly/4pLUotH

Follow us for more

28/08/2026

As Meta is projected to edge out Google as the world’s largest advertising platform with US$243 billion in expected revenue, AI-fuelled algorithms are driving digital ad matching to unprecedented levels. However, because competing platforms and advertisers rely on similar machine learning techniques and public data, ads often end up targeting the exact same people at the same time.

A new study co-authored by Professor Jesse Yao, Associate Professor in the Department of Marketing at CUHK Business School, applies game theory to reveal how businesses can navigate this strategic dilemma:

📌 Focusing on high-intent buyers triggers expensive head-to-head auctions against competitors, while broad targeting risks wasting resources on uninterested audiences.
📌 Data privacy regulations prevent algorithms from achieving 100% accurate targeting, meaning perfect prediction remains out of reach.
📌 Wise advertisers strategically target moderately interested audiences where competitor clutter is lower, securing conversions without costly bidding wars.
📌 Using unique, first-party data reduces overlap with rivals and creates a sustainable competitive advantage.

Read the full article: https://bit.ly/4xlyBMz

Follow us for more

28/08/2026

The global race for AI leadership is intense, but data compliance concerns are reshaping how businesses choose their technology stack. A new study co-authored by Professor Jiang Wenxi, Professor of Finance, and Professor Gao Zhenyu, Associate Professor of the Department of Finance at CUHK Business School, reveals that data security is the primary driver behind AI adoption strategies among Chinese listed firms.

📌 Researchers discovered a dramatic spike in the adoption of domestic models like DeepSeek compared to foreign GenAI by examining interactions between firms and their investors on stock exchange platforms.
📌 Strict data regulations, including the Cybersecurity Law, the Data Security Law and the Personal Information Protection Law, require sensitive data to be stored locally, making data transfers to foreign servers pose a compliance risk.
📌 Chinese firms strategically limit foreign AI to non-critical tasks like customer service, basic translation, while shareholders actively reward strategic-sector companies that adopt domestic AI solutions.

Read the full article: https://bit.ly/3SbdYmY

Follow us for more

28/08/2026

The foundational data used to train AI models may carry the imprint of societal biases, raising questions about whether these powerful technologies simply echo existing gender stereotypes. A new study co-authored by Professor Li Hongfei, Assistant Professor in the Department of Decisions, Operations and Technology at CUHK Business School, reveals that large language models exhibit a largely balanced perception of both masculine and feminine traits, until they were pushed to think technically. A systematic bias quietly returns once the model is tasked with formulaic investment calculations.

📌 When tasked with describing successful entrepreneurs or evaluating pitches, ChatGPT avoids gender stereotypes and even gives higher marks to collaborative, people-focused, and warm tones.
📌 When instructed to act as a venture capitalist using mathematical formulas to analyse investment opportunities, AI treats the task as a simple math exercise, bypassing its ethical guardrails and favouring risk-embracing, stereotypically masculine traits.
📌 As AI models are trained on extensive data, human-in-the-loop oversight is critical when using algorithmic systems to screen pitches, ensuring that technology promotes fairness rather than reinforcing outdated biases.

Read the full article: https://bit.ly/3TPosss

Follow us for more

24/08/2026

Smarter isn’t always better. The most advanced AI can overthink and become more irrational, while simpler models make near-perfect decisions in key business scenarios. It’s an AI paradox!

“LLMs don’t just mirror human biases, but also often amplify them,” says Zhi Chen, Associate Professor in the Department of Decisions, Operations and Technology at CUHK Business School. “Our experiments show that LLMs consistently replicate the classic too-low or too-high ordering bias well-documented in humans, ordering too little in high-margin scenarios and too much in low-margin ones.”

Professor Chen’s study also finds that an AI overreacts to market trends 70% more than a human would, leading to bigger mistakes. This means we need to be smart about how we use AI: choose the right tool for the job, and always keep human expertise in the loop to guide these powerful systems.

Read more: https://cbk.bschool.cuhk.edu.hk/does-smarter-ai-generate-more-human-errors/

Businesses hand more and more decisions to AI, but Professor Chen Zhi finds some of these tools can overthink simple problems and make worse choices than humans

17/08/2026

The rise of AI-powered targeting has revolutionised digital marketing. As companies increasingly adopt these sophisticated techniques, is simply having a good algorithm enough?

“When competition is strong, companies have a high chance of targeting the same pool of individuals, especially if their algorithms use similar mechanisms,” says Jesse Yao, an Associate Professor at the Department of Marketing at CUHK Business School.

His study reveals that firms that prioritise highly interested consumers may find their targets shared with rivals, but casting a wider net to a broader target may risk wasted ad spend. However, building unique algorithms may benefit rivals by reducing competition, making it easier for rivals to access high-quality targets.

Read more: https://cbk.bschool.cuhk.edu.hk/if-everyone-uses-ai-who-stands-out/

When competition is strong, companies have a high chance to target the same individuals, especially if their algorithms are similar , Professor Jesse Yao finds