BOCHK Challenge 2021-22

Background:
The BOCHK Challenge, organized by Bank of China (Hong Kong), aims to explore the potentials , nurture talents of youngsters and create groundbreaking business models with innovative technological solutions. The theme of the challenge this year covered five of the most trending technologies in the market. By applying these technologies to the financial industry, students strived to improve customer experience and operational efficiency.

This year, more than 144 teams of college students and start-ups submitted their proposals. Teams were invited to attend workshops conducted by Deloitte and Alibaba about trends in financial technology and innovation. Shortlisted teams which were mentored by BOCHK, Deloitte and Alibaba experts competed for 7 awards and the winning teams were awarded HK$ 590,000 prizes in total.

Website / Image URL:
https://www.bochk.com/dam/more/bochkchallenge/2021/en.html

Award(s):
Gold Award for Innovative Design

Awardee(s):

Team name: Maldin Consulting

Miss Lai Hiu Chi Emily, BBA(IBGM), Year 1

Miss Wong Tsz Kiu, BBA(Law)&LLB, Year 1

Students Sharing:
“Maldin Consulting is honoured to stand out among 144 teams of college students and start-ups to win the Gold Award for Innovative Design in the “BOCHK Challenge 2021-22”. We hope that our AI financial advisor can be a breakthrough from the traditional industry framework, enhancing the operational efficiency of the bank. We aspire to provide the right service, to the right customers, at the right time.

We would like to thank the organiser, Bank of China (Hong Kong) (“BOCHK”), the Hong Kong Monetary Authority, Hong Kong Science and Technology Parks Corporation, Hong Kong Productivity Council, Deloitte China and Alibaba Cloud Intelligence for hosting this competition. We would also like to sincerely thank all BOCHK mentors and co-organizer experts, Garry for providing us with guidance on business innovation, financial technology applications, technology, presentation skills, etc. Thank you Gary for enlightening us with his perspective on digital solutions: “Banking is and always will be a business established by relationships. The purpose of technology in the end is to help humans serve humans better.”

Finally, I have to thank each and every one of my team members for their hard work. We would not be able to reach so far without any one of you. Your work ethic and persistence motivate me to always deliver my best.

This marks the end of our first year of undergraduate, we look forward to challenging ourselves in the years to come!”

(by Lai Hiu Chi Emily)

“My journey in the BOCHK Challenge had been extremely rewarding. I am immensely grateful for having some of the most amazing people as my teammates, Emily, Carina, Megan and CY. Their attitudes, talents and professional skills truly inspired me. We had worked days and nights striving for perfection. Even though we were only a group of year 1 students, our hard work eventually brought us success.

We were very fortunate to have the opportunity to learn from our mentor, Garry from Alibaba Cloud. He enlightened us by sharing the latest industry trends and telling us the essence of asset management is relationship building. At the end of the day, we as humans crave to be emotionally connected. The ultimate purpose of technology in business is to establish a deep and long-lasting relationship with the customers.”

(by Wong Tsz Kiu Jasmine)

Other Events
港大经管学院最新多模态AI图像生成能力排名出炉 部份中国人工智能模型表现突出
2025 | 学院新闻
港大经管学院最新多模态AI图像生成能力排名出炉 部份中国人工智能模型表现突出
港大经管学院今日发表《人工智能模型图像生成能力综合评测报告》,针对15个“文生图模型”及7个“多模态大语言模型”进行全面评估。研究显示,字节跳动的即梦AI和豆包,以及百度的文心一言,在新图像生成的内容质素及图像修改的表现突出;而早前引起全球关注的DeepSeek最新推出的文生图模型Janus-Pro,则在新图像生成方面表现欠佳。研究亦发现部分文生图模型虽然在内容质素方面表现优异,却在安全与责任方面的表现强差人意。整体而言,与文生图模型相比,多模态大语言模型整体表现较佳。
人工智能图像生成评测成绩单公布: 字节跳动百度表现亮眼,DeepSeek Janus-Pro表现欠佳
2025 | 研究
人工智能图像生成评测成绩单公布: 字节跳动百度表现亮眼,DeepSeek Janus-Pro表现欠佳
如今,人工智能领域的前沿模型技术已经从文本处理拓展至视觉信息的深度理解与生成。这些模型既能精准解读图像语义,又能根据文字描述创作出兼具真实感与艺术性的视觉内容,展现出令人惊叹的跨模态理解与创作能力。本研究聚焦全新图像的生成和基于现有图像的图像修改两大核心任务,提出了一套系统性的人工智能模型图像生成能力评测框架。我们基于多维测试集的构建与专家评审,对15个专业文生图模型和7个多模态大语言模型的图像生成能力进行了全面评估。结果显示,字节跳动的即梦AI和豆包以及百度的文心一言在新图像生成的内容质量与修改任务中表现突出,位列第一梯队。对比不同类型的AI模型,我们发现,相对于专业文生图模型,多模态大语言模型整体表现更佳。