Hitachi Vantara’s Lee Nolan on Responsible AI Success
Lee Nolan, GM UK&I at Hitachi Vantara, explains why AI agents need guardrails, an ethics code and senior ownership before they reach real data or customers
TL;DR
- AI agents should not have unrestricted access to organizational systems, mirroring human access limitations.
- Guardrails are essential to protect personal and sensitive information from AI agents.
- Deployment without a clear business case and measurable value is considered irresponsible AI.
- Over-reliance on AI could lead to a loss of organizational identity and differentiation.
- Companies need an internal AI code of ethics to guide agent behavior and decision-making.
- AI deployment is a leadership issue, involving strategic, risk appetite, and values decisions.
- AI experimentation should be freely encouraged in sandboxed environments, with strict control over deployment.
- Physical AI should be treated as a safety-critical engineering project, not just an AI project with a motor.
- Safety engineering principles like permits to work and fail-safe defaults are crucial for physical AI.
- Hitachi Vantara highlights significant global AI investment waste due to weak data governance.