【会议演讲PPT】Gartner+生成式+AI+对治理和政策的影响-34页_2mb

报告摘要

Gartner Webinar Summary: Generative AI Impact on Governance and Policy

Overview

This Gartner webinar explores the impact of Generative AI (GenAI) on organizational governance and policy, drawing from analyst expertise and data. It addresses risks, benefits, and recommendations for integrating GenAI in a structured manner, emphasizing the need for balanced use cases and proactive governance.

Key Statistics and Findings

GenAI Adoption: Approximately 170 million people used ChatGPT in its first two months (Business Insider), with nearly 70% keeping its use secret from bosses.
Enterprise Policies: Despite 68% of respondents believing benefits outweigh risks, nearly 1 in 3 enterprises prohibit GenAI use, and more than 1 in 3 lack guidelines, based on Gartner polls.
Risk Perception: Organizations face various risks, including unreliable outputs, data privacy issues, and regulatory challenges.

Understanding Generative AI (GenAI)

Definition: GenAI encompasses AI techniques that generate new artifacts by learning from existing data, including large language models (LLMs) like ChatGPT.
Use Cases: Appear in various applications, such as legal research, software engineering, and customer service.

Risks and Benefits

Benefits: Generally outweigh risks for most organizations (68% agreement), with specific applications in innovation and efficiency.
Risks: Include unreliability, data breaches, bias, intellectual property theft, cybersecurity threats, and societal impacts like job displacement and misinformation.
Governance Importance: Risks must be assessed and integrated into governance models to balance use cases and policy guidance, as part of an evolving risk landscape.

Policy and Governance Recommendations

Develop an AI Policy: Define GenAI, acceptable use cases, user obligations, and monitoring mechanisms.
Governance Framework: Combine policy with controls, such as data loss prevention (DLP) tools and IT partnerships.
Structured Approach: Use factors like risk assessment and acceptable use cases to guide decision-making.

Deployment Models

Public vs. Industry-Tailored:
- Public Models: Accessible, low-cost options like ChatGPT but with higher risks of hallucinations.
- Industry Models: Customizable and lower-risk, requiring stronger internal capabilities for integration.

Conclusion

Organizations should focus on understanding business-specific GenAI use cases, mitigating risks, and gaining executive buy-in. Gartner supports this through research, tools, and resources for informed policy creation.