AI Guide For Boards in Singapore

AI GUIDE FOR BOARDS IN SINGAPORE

Section II: Strategy and Value Creation

AI Performance Indicators: Examples

Examples of AI-specic key indicators include: • Percentage of workows augmented by AI. • Percentage of employees actively using approved AI tools.

• Human review burden per AI-assisted process. • AI containment rate in customer operations. • Model dri frequency. • Prompt-to-output success rates. • AI-generated defect rates. • Average time saved per workow. • Percentage of pilots converted to production. • AI-related risk incidents or policy breaches.

Boards should also be cautious of relying on activity or adoption metrics as standalone evidence of value. ese indicators can be useful in the early stages of AI adoption, particularly to assess awareness, experimentation and initial usage. However, as AI initiatives mature, they should be complemented by outcome-based measures that show whether adoption is translating into operational, nancial, customer or risk- management impact.

Examples of potentially weak or misleading indicator include: • Number of AI pilots launched. • Number of licences purchased. • Number of AI training classes conducted. • Volume of prompts generated. • Total chatbot interactions without outcome measures. • Generic productivity claims without operational evidence.

Beyond benchmarking, the measurement framework should indicate if the AI initiatives can be scaled beyond pilot environments. indicators, such as workow improvements, eciency gains, reduction in errors, improved customer experience and operational impact. Boards should ensure that adoption metrics are complemented by outcome-based High usage levels may indicate curiosity or experimentation without necessarily improving outcomes, productivity or eciency.

Boards should focus on whether AI initiatives are producing measurable business outcomes rather than technology activity alone.

5.1.4 Adoption as an indicator of value

Adoption rates can sometimes provide an early signal of value, particularly in the initial stages of deployment. However, adoption should be treated as an indicator of potential value, rather than evidence of business value in itself.

45

Made with FlippingBook Digital Proposal Creator