The AI Development Life Cycle Revisited
Last reviewed:
AI development mirrors the software life cycle plus a data obsession and continuous monitoring. Policies, procedures, best practices and ethics apply at every phase, and TEVV is continuous across the whole life cycle.
AI development mirrors the software life cycle, plus a data obsession and continuous monitoring. Policies, procedures, best practices and ethics apply at every phase.
Planning and design → data collection and preparation → model development (selection and training) → model testing and evaluation → deployment → monitoring and maintenance → decommissioning.
TEVV is continuous → testing, evaluation, verification and validation run across the whole life cycle, not at one gate. A pre-deployment pilot is a late-stage validation activity under realistic conditions, sitting immediately before go-live.
Key terms - quick answers
What is “TEVV”?
What is “Pre-deployment pilot”?
Keep going - free
Every study note on this site is free. So are these: