Bias in AI Systems
📌 A viral example from seven years ago involved Google Photos confusingly tagging African American individuals as gorillas.
📊 A study by MIT showed significant differences in facial recognition system biases across various communities.
🌐 These biases are not exclusive to image recognition, also appearing in automatic translators through detectable gender prejudices.

Data Management and Legal Frameworks
🧠 Artificial Intelligence requires large volumes of data for training predictive models, making data use and management a delicate issue.
🛡️ The primary regulation governing data in IT is the Organic Law on Data Protection and Digital Rights (referencing GDPR principles).
📜 Key professional roles in data management include the Data Controller (responsible for defining purpose and means), the Data Processor (acting only on documented instructions), and the Data Protection Officer (DPO) (a specialist, often mandatory for public administrations or high-volume/sensitive data processing).

User Rights and Ethical Considerations
🔒 Article 5.1(f) of the GDPR mandates a duty of confidentiality and security, ensuring adequate protection against unauthorized processing, loss, destruction, or accidental damage.
⚖️ A crucial principle often unmet, as seen in the Google case, is the veracity and accuracy of data—ensuring data is updated and truthful.
🚫 Article 22 of the GDPR grants users the right not to be subject to automated decisions evaluating personal aspects (performance, health, preferences, etc.), unless explicit consent is given.
💡 The speaker defines ethics not just as a list of prohibitions, but as the guiding principles for what we want to promote as a collective.

Key Points & Insights
➡️ Key ethical foundations for computer scientists include preventing harm, avoiding informational inequality and injustice, and non-interference with moral autonomy.
😟 Users must be highly cautious about what they upload online, as "what you share" now defines you, potentially impacting professional careers.
✅ When profiles are built (even with consent), controllers must use proper procedures to ensure accuracy and reduce the risk of discriminatory effects based on factors like race or political opinion.

📸 Video summarized with SummaryTube.com on Oct 04, 2025, 05:06 UTC