2025年人工智能发展态势报告_智能体_创新与转型_32页_8mb

Summary of McKinsey AI Report 2025

• Overview: According to the report, most organizations are using AI, but scaling it for enterprise-level impact remains challenging. Nearly all survey respondents report regular AI use in at least one business function, yet two-thirds are still in the experimentation or piloting phase, with only one-third beginning to scale AI across the enterprise. This highlights a gap between AI adoption and deep integration for material benefits.

• Adoption and Scaling:

  • AI use is expanding across various business functions, with companies increasingly deploying it in areas like IT, knowledge management, marketing, and software engineering.
  • Larger organizations, particularly those with revenues over $5 billion, are more advanced in scaling AI compared to smaller ones, with nearly half of their respondents reaching the scaling phase, versus 29% of smaller firms.
  • Foundational models for agentic AI are being explored, but scaling is limited, with only about one-third of organizations using AI agents, and even less scaling in specific functions.

• AI Agents:

  • AGI agents are gaining attention, with 62% of organizations at least experimenting with them. Leadership commentary notes that while excitement is high, widespread adoption is not yet achieved; AI high performers lead in agent scaling, using them in key functions like IT and knowledge management.
  • Despite hype, most organizations are still in early stages, often focusing on pilot projects rather than full integration.

• Impact on Business:

  • Respondents report cost savings in functions such as software engineering and manufacturing, but enterprise-wide EBIT impact is limited. Only 39% attribute any EBIT effect to AI, with high performers achieving significant value through transformative innovation, productivity gains, and redesigning workflows.
  • AI drives innovation and customer satisfaction for many, with high performers setting objectives beyond efficiency, such as growth and transformation. However, financial benefits at scale are rare.

• High Performers:

  • These organizations (about 6% of respondents) are three times more likely than others to fundamentally redesign workflows and set bold AI agendas for transformation. They invest more in AI budgets, exceeding 20% of digital spending, and engage senior leaders who champion AI initiatives.
  • High performers also excel in management practices including human-in-the-loop validation, agile delivery, robust data infrastructure, and comprehensive risk mitigation.

• Workforce Impact:

  • Expectations on employment vary, with a plurality predicting no change or slight increases in workforce size, while 32% anticipate decreases. Larger companies show more pronounced changes.
  • On the positive side, organizations are hiring for AI roles such as data scientists and software engineers, with larger firms leading in these efforts.

• Risks and Mitigation:

  • AI-related risks like inaccuracy and explainability are common, with 30% experiencing negative consequences; high performers face more risks but mitigate them more effectively. Management practices for strategy, talent, data, and governance are improving, with organizations increasingly addressing ethical and regulatory challenges.

• Conclusion and Key Takeaways:

  • AI adoption is broad, but scaling requires a transformative approach focused on innovation, not just efficiency. The report suggests that high performers prioritize workflow redesign, leadership commitment, and comprehensive management practices. Enterprise-wide impact remains elusive, but as AI tools like agentic systems evolve, opportunities for deeper integration and value capture grow. Lessons emphasize the need for ambitious agendas, robust risk management, and a focus on hybrid human-AI collaboration to navigate AI's evolving landscape and achieve sustainable success.