Position Bias in AI Models Threatens Accuracy in High-Stakes Applications, MIT Warns
Attention bias in AI models favors text beginnings, ends, MIT study maps root, proposes mitigation strategies.
Read the original at digitalinformationworld.com ↗
- Topics
- Bias & Fairness
- Mentions
- MIT
- Tracked via
- “AI Bias”