Page 163 - AI Ver 3.0 class 10_Flipbook
P. 163

• Non-maleficence (Do Not Harm): This principle focuses on avoiding actions that could harm others whether
                   intentional or unintentional to an individual or a community.
                    • Justice: This principle ensures fairness in distributing healthcare resources, treatments, and opportunities. It
                   emphasises on equality and avoiding discrimination in medical decision-making.



                                 Brainy Fact



                        Maleficence refers to the deliberate act of causing harm, injury, or wrongdoing to others.





                 CASE STUDY: AI for Early Detection of Mental Health Issues

                 Background

                 A healthcare organisation used an AI system to help mental health of professionals. Identify people at high
                 risk for mental health problems like depression and anxiety. The goal was to spot these issues early, improve
                 patient  care,  and  use  resources  more  effectively.  The  AI  analysed  various  data  sources,  including  medical
                 records,  demographics,  social  media  activity,  and  behavioural  patterns,  to  predict  the  likelihood  of  mental
                 health disorders.

                 However,  the  system  introduced  unintended  consequences,  resulting  in  biased  treatment  towards  specific
                 patient groups.
                 The Problem It Caused

                 The  AI  system  incorrectly  flagged  women  from  low-income  communities  as  having  a  higher  risk  of  mental
                 health issues based on factors like their social media activity and signs of financial stress. However, many of
                 these women were not experiencing mental health disorders but were dealing with financial difficulties and
                 caregiving responsibilities.

                 Conversely, the system failed to identify individuals from wealthier backgrounds who might also be at risk but
                 did not exhibit obvious signs of stress, leading to missed opportunities for an early intervention.
                 Why the Problem Happened?

                    • Bias in data: The AI model was trained on datasets that primarily represented wealthier individuals with better
                   access to healthcare. Social media data further amplified the issue, as people from lower-income communities
                   often have limited access to online mental health resources.
                    • Overemphasis on social media: The system relied excessively on social media activity (like post frequency and
                   tone) to predict mental health without considering how social or economic stress factors can affect someone's
                   mental health.

                    • Ignoring social factors: The AI didn't take into account the broader social issues, like poverty or lack of access
                   to care, which can significantly affect mental health.
                 The Ethical Problems

                    • Bias: The AI overestimated the mental health risks for low-income communities while overlooking others who
                   needed care. This exacerbated existing healthcare inequalities.


                                                            Revisiting AI Project Cycle & Ethical Frameworks for AI   161
   158   159   160   161   162   163   164   165   166   167   168