AI ETHICS
EU Ethics Guidelines for Trustworthy AI
AI ETHICS FOCUS
Control and oversight measures through AI impact analysis (economic, societal, legal, and ethical).
Human in the loop, Human on the loop, Human in command.
Technical robustness and safety through error handling, data security, fallback plans. Decisions must be accurate and their outcomes reproducible.
Safety- and security-by-design approach to show the solution is verifiably safe. Clarify and assess the potential risks associated with the system.
Data privacy and protection at all stages of the system lifecycle. Training dataset quality must be high and free from biases, errors, inaccuracies, and mistakes.
Processes and datasets must be tested and documented during planning, training, testing, and deployment.
Access to data must be governed and controlled.
Transparency through traceable decisions via logs, explainable decision-making.
Communication of the systems capabilities and limitations to the users.
Societal and environmental well-being through impact analysis and the change of social skills of workers.
This will be considered during the design and development of the didactic concept.
Accountability through auditability of the solution.
Documentation of potential negative impacts, readress mechanisms, and assess specific legal barriers.