BBA(IBGM) student Singh Shivang wins the regional final of Hult Prize 2021 Challenge

BBA(IBGM) student Singh Shivang wins the regional final of Hult Prize 2021 Challenge

Congratulations to Shivang Singh (BBA(IBGM), Year 4) who was a member of the winning team “Edilery” in the Hult Prize 2021 Challenge – Regional Final in Manila Impact Summit.

Proudly acclaimed as the “Nobel Prize for Students”, the Hult Prize is the world’s largest social entrepreneurship competition organised by the Hult International Business School and the Union Nations Foundation. The theme of 2021 was “Food for Good – transforming food into a vehicle for change”.

Student teams are required to solve a pressing social issue by developing a scalable and sustainable social enterprise. They were asked to pitch ideas on building food enterprises that will impact the lives of 10 million people by 2030 while strengthening communities, increasing incomes, feeding the hungry and creating jobs.

The team “Edilery” came up with an idea that aspired to drive sustainable impact in fighting climate change, malnutrition and gender inequality. With edible cutlery as the vehicle of change, they aimed to create livelihoods to support communities while eradicating plastic cutlery from the global food and beverage industry and promoting nutrition in regular dietary intake.

Winners of the Regional Finals will join the 2021 Hult Prize Accelerator Programme in the United Kingdom with training, coaching and networking opportunities. Selected teams from the Accelerator will pitch at the Hult Prize Annual Awards Ceremony and Gala to be hosted at United Nations Headquarters in New York in September 2021. One student team will be selected as Hult Prize Laureate and win US$1,000,000 to launch their enterprise.

Learn more: https://www.hultprize.org

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。