Model Guardian

Machine Learning [ML] models bring unprecedented accuracy and efficacy to challengingly complex optimization problems. In contrast to rules-based models, ML models derive the rules directly from the data… the more, the better generally). While the foundation in data spares ML models from risks around assumptions and other pre-conceived notions, it does not make them invulnerable.

If built on non-representative data, ML models can perpetuate human prejudice just as the human-designed rules can skew classical models. Unintended biases can be included into these due to patterns in the data. Just as AI gains its strength from harvesting deep and wide datasets, a poor dataset will condemn an AI to inadequacy, risking to mislead rather than to inform. Skewed data injects subjectivity into an otherwise objective decision making process.

Key Features of the Tool

  • Enable Ethical AI
  • Ensure Compliance
  • Assist with Risk Management
  • Identify blind spots and opportunities in the data

Back to Africa TMT Lighthouse page

View now
Did you find this useful?