**Technical AI Ethics Challenge:**

Technical AI Ethics Challenge:

Title: "Fairness in Autonomous Decision-Making with Limited Feedback"

Context: A city has deployed an autonomous traffic management system that uses machine learning to optimize traffic signal timings. The system is designed to minimize travel times and reduce congestion. However, the AI model is trained on historical data that may contain biases related to socioeconomic factors such as income, education, and employment status.

Constraints:

1. The city's resources are limited, and the AI system can only receive feedback in the form of traffic data (e.g., speed, volume, and delay) every 15 minutes.
2. The city has a strict policy of not revealing individual-level data to prevent potential misuse.
3. The AI system must maintain fairness and avoid perpetuating existing biases, particularly those related to socioeconomic factors.

Challenge: Design an AI system that can:

1. Learn from limited feedback and adapt to changing traffic patterns.
2. Detect and address biases in the training data without access to individual-level data.
3. Ensure fair treatment of all road users, without sacrificing the overall performance of the system (e.g., minimizing travel times and reducing congestion).

Evaluation Criteria: The proposed solution will be evaluated based on the following metrics:

1. Bias reduction: The proposed solution should demonstrate a significant reduction in the bias present in the initial training data.
2. Performance: The proposed solution should maintain or improve the overall performance of the system (e.g., minimizing travel times and reducing congestion).
3. Fairness: The proposed solution should ensure fair treatment of all road users, without sacrificing system performance.

Note: The challenge requires the design of an end-to-end system that integrates data collection, model training, and adaptation. A detailed technical report outlining the architecture, methodology, and experimental results is expected as part of the submission.

Submission Guidelines:

• Please submit a detailed technical report (max. 10 pages) outlining the proposed solution, methodology, and experimental results.
• Include a clear explanation of the system architecture, model selection, and adaptation mechanisms.
• Provide a detailed analysis of the evaluation metrics and results.
• Include a concise summary of the proposed solution, highlighting its key contributions to fairness and performance.

Deadline: February 28, 2026.

---

Publicado automáticamente