本报告探讨了⼈⼯智能的当前格局，审视其巨⼤潜⼒以及需解决的重⼤⻛险。我们深⼊讨论政府在管理这⼀强⼤技术⽅⾯的⻆⾊，重点关注在利⽤其优势的同时减轻其弊端的策略。确保⼈⼯智能成为促进经济增⻓、社会进步和更可持续未来的推动⼒所需的全⾯⽅法和有效政策和法规⾄关重要。

The global artificial intelligence (AI) revolution is here and companies across Asia Pacific (APAC) are gearing up to maximize their AI advantage. If enterprises want to capitalize on new initiatives, IT leaders need to harness the right digital infrastructure strategy to ensure AI and other emerging technologies are supported from the very foundational layers. This includes having the right data infrastructure, systems, and IT environments to future proof their AI investments.

本报告聚焦 AI Agent 技术对药企研发的赋能。Deepseek 凭借推理能力突破、开源等优势引爆行业，推动 AI Agent 从工具向 “数字员工” 进化，具备自主决策等特性。全球 AI Agent 市场 2024 年规模约 51 亿美元，2030 年有望达 471 亿美元。智慧芽即将上线多个生物医药场景 AI Agent，可助力专利撰写、临床分析等，释放药企创新潜力，提升研发效率。

能够制造并使用工具成为人类进化史上一道显著的分水岭，而当下如何更好的使用AI工具已然 成为人类在产业应用、生产生活与学习工作中的热门议题。随着大模型、生成式AI技术的到来，其 强大的数据处理、学习泛化与内容生成能力，高质效加速了各行各业人工智能技术的赋能进程，为 AI可赋能的场景领域、扮演角色提供更多创新性与可能性。人工智能应用正加速扩散，渗透到办公、 设计、传媒、法律、游戏、教育、汽车等多领域。

AI 大模型与人形机器人融合：随着人工智能算法和机器学习技术的进步，AI 大模型等人工智能技术正转变机器人的决策逻辑，帮助人形机器人软件层面开始具备较强的解决方案，使其变得越来越智能。这些技术将使人形机器人在动态环境中更好地感知、推理和行动，理解人类语言，并从经验中学习。

In the financial services industry, GenA.I. applications have the potential to lead to more curated customer experience, and more efficient ways of processing and utilising digital information. However, the adoption of GenA.I. could also give rise to new risks and challenges. This suggests that amidst accelerated GenA.I. innovation, its adoption requires a critical focus on safety, trust and integrity. In Hong Kong, authorities have begun to update regulatory guidelines and to launch initiatives in support of responsible GenA.I. adoption and innovation.

The report highlights three key AI-driven levers: optimizing direct procurement by using AI to track raw material costs and analyze supplier performance for improved negotiation outcomes; enhancing process efficiency through generative AI to automate marketing content creation and campaign planning, reducing labor costs; and unlocking the potential of indirect procurement by leveraging AI to enhance spend transparency, consolidate demand, and optimize supplier management.

尽管业界对生成式 AI 充满期待，但其在内容供应链（CSC） 中的落地速度未达预期。截至 2024 年底，仅 50% 的组 织达成了 AI 采用目标，仍有较大差距。 尽管挑战重重，组织对生成式 AI 的信心却持续攀升。相 比去年，近三分之二（64%）的组织对生成式 AI 的影响 力更有信心。84% 的组织仍坚信，生成式 AI 能够高效驱 动个性化体验的规模化生产与交付。

This report focuses on the AI opportunities in the automotive industry in 2025, pointing out that AI is evolving from generative to agent-based forms. While its implementation has been slower than expected, its potential remains significant. AI can drive technological innovations such as software-defined vehicles and autonomous driving, and has application scenarios in multiple links of the value chain, including research and development (R&D), production, and beyond. It is expected to bring a 40-60% increase in marginal profits.

基于AI的生态系统能从互联的传感器和控制器中提取更多信息，有可能使每个参与设备的价 值大幅超越其基本功能。然而，每个设备的AI贡献价值会因使用案例和解决方案规模而异， 所以AI并不能成为随意提高设备价格的理由。功能价值仍然设定了客户对成本的基本预期， 因此，开发盈利的AIoT设备的关键在于尽可能高效地添加AI功能和生态系统连接——而具有 成本效益的连接正是本文的重点。