Challenge / AI for Good
A good intention does not remove the need to test harm, bias and failure.
Teams propose a bounded use of AI, examine who is affected and test whether a simpler non-AI approach may serve the need better.
The challenge should follow, not replace, learning about data, models and evaluation.OPENQUESTION
01 / Question02 / Understand03 / Build04 / Verify01
BIAA / PROBLEM
Begin with people and context
Teams define the need, stakeholders and decision being supported before selecting a model or tool. The challenge rewards problem clarity rather than novelty alone.
02
BIAA / AUDIT
Audit data and failure
The evidence pack identifies data origin, missing perspectives, likely errors, privacy risks and the role of human review. Sensitive or high-impact use cases should be excluded from classroom implementation.
03
BIAA / ALTERNATIVE
Compare an AI and non-AI route
A rules-based, manual or educational intervention may be safer and more effective. Teams explain why AI is—or is not—justified and how the choice will be monitored.
BIAA / NEXT STEP
Build AI foundations Build AI foundations
The challenge should follow, not replace, learning about data, models and evaluation.