IDC PeerScape: Practices for Managing Safe AI

Description

This IDC PeerScape explores how some organizations are educating employees, vendors, and partners on what safe AI is all about, why it's important, and how to maintain successful safe AI practices in a constantly evolving AI landscape."It is well understood that AI makes frequent errors - or 'hallucinations' - due to inaccurate, misleading, or incomplete data," says David Weldon, research adjunct in IDC's IT Executive Programs (IEP). "Because AI has little ability to distinguish good data from bad, it can be difficult for businesses and citizens alike to feel such lapses might not do genuine harm to people or organizations. That is where safe AI comes into play, enabling workers and organizations to place trust and confidence in their AI investments."

Product Code: US53799325

IDC PeerScape Figure

Executive Summary

Peer Insights

Practice 1: Provide Training to Help Staff Understand AI's Capabilities and Limitations, and Ensure AI Vendors Meet Standards for Data Privacy, Security, and Ethical Use
- Challenge
- Examples
  - Fox Rothschild
  - Carnegie Mellon University
  - Z/Yen Group
- Guidance
Practice 2: Take a Proactive and Structured Approach to AI Adoption That Ties Ethical Considerations to Business Gains
- Challenge
- Examples
  - Fox Rothschild
  - Carnegie Mellon University
  - Z/Yen Group
- Guidance
Practice 3: Invest in AI Tools and Practices That Enhance Explainability and Documentation to Aid in Building Understanding and Confidence
- Challenge
- Examples
  - Fox Rothschild
  - Carnegie Mellon University
  - Z/Yen Group
- Guidance
Practice 4: Test and Evaluate AI Systems Regularly for Compliance, Bias, and Security Vulnerabilities
- Challenge
- Examples
  - Fox Rothschild
  - Carnegie Mellon University
  - Z/Yen Group
- Guidance