Page 164 - AI Ver 3.0 class 10_Flipbook
P. 164

• Inaccuracy: If the AI makes wrong predictions, it could lead to unnecessary treatments for some individuals
                 while others at risk went unnoticed, compromising the system's reliability.

                 • Privacy Concerns: Using social media activity for risk predictions raised concerns about people's privacy and
                 whether they were properly informed about how their data was being used.
              Using Bioethics to Fix the Problem

              By applying four principles of bioethics, we can improve fairness and effectiveness of AI system:

                 • Respect for Autonomy
                 o  Transparency and consent: Patients must be informed about how their data is used and given the choice

                    to opt out of having their personal or social media data analysed.
                 o  Clear communication: Mental health professionals should clearly explain how the AI system functions, its
                    data sources, and its role in patient care.

                 • Non-maleficence (Do Not Harm)

                 o  Minimise harm:  The  AI  should  undergo  rigorous  testing  to  ensure  it  does  not  disproportionately
                    misclassify or harm specific groups, such as low-income individuals.

                 o  Human oversight: Predictions made by the AI should be reviewed by mental health professionals to
                    mitigate potential harm.
                 • Beneficence (Maximum Benefits)

                 o  Benefit to all: The system should be retrained to assist all patients, especially those traditionally overlooked
                    or underserved, by including diverse data from varied socio-economic, racial, and demographic groups.

                 o  Data diversity: The training dataset must be expanded to better represent all populations and improve
                    the system's accuracy.

                 • Justice

                 o  Fairness: The AI should treat all groups equitably and avoid reinforcing existing disparities.

                 o  Address bias: Developers must critically assess the dataset and address any systemic biases, such as those
                                                                                                                                                                                         Exercise
                    related to race, gender, or class, in the AI model.
              Proposed Solutions

                 • Improved data collection: Gather data from a wide range of individuals across different socio-economic and                                                        Solved Questions
                 racial backgrounds to build a more inclusive and balanced model.
                                                                                                                                                                             SECTION A (Objective Type Questions)
                 • Bias detection and mitigation: Implement advanced techniques to identify and eliminate biases in the dataset
                                                                                                                                                  uiz
                 and the AI's algorithms.
                                                                                                                                            A.  Tick ( ) the correct option.
                 • Human review: Mental health professionals should actively monitor and validate AI predictions to ensure
                                                                                                                                                1.  What is the primary purpose of AI Project Cycle?
                 accuracy and fairness.
                                                                                                                                                  a.  To provide a set of random steps for AI development
                 • Privacy protections: Ensure robust data security and privacy measures for personal data, especially from social
                                                                                                                                                  b.  To systematically plan, develop, and deploy AI solutions
                 media. Patients should also have the ability to withdraw consent for data usage.
                                                                                                                                                  c. To focus only on the final implementation of AI systems
              By applying these bioethics principles, the AI system can be made more accurate, fair, and respectful of patient
                                                                                                                                                  d. To perform data analysis without a structured plan
              privacy. This will help ensure that the system benefits everyone, regardless of their background.

                    162     Touchpad Artificial Intelligence (Ver. 3.0)-X
   159   160   161   162   163   164   165   166   167   168   169