Bias and data privacy

AI for MEL: Tutor-Led

[vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h2″ text_direction=”default”]Bias and data privacy[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Bias[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Bias refers to systematic and unfair discrimination that results in unequal outcomes across individuals or groups.

In AI systems, algorithmic bias arises when the outputs of a machine learning model systematically disadvantage certain populations due to skewed data, flawed assumptions, or structural inequalities embedded in the design process.[/nectar_responsive_text][image_with_animation image_url=”4390″ image_size=”full” max_width=”100%” max_width_mobile=”default” animation_type=”entrance” animation=”None” animation_movement_type=”transform_y” hover_animation=”none” alignment=”center” border_radius=”none” box_shadow=”none” image_loading=”default” display_title=”1″ fit_to_container=”1″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Algorithms are not neutral; they reflect the values, limitations, and contexts of their developers and data sources.

“Algorithmic systems can reproduce and magnify the discriminatory effects of historical social inequalities embedded in the data or decision logic.” (Barocas & Selbst, 2016)[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Types of Bias in Ai[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]

Selection Bias

– Training data is not representative of the real population.
– Example: Facial recognition struggles with darker skin tones if mostly trained on lighter-skinned faces, causing discrimination.

Confirmation Bias

– AI system is overly reliant on pre-existing patterns in the data, reinforcing historical prejudices.
– Example: Hiring algorithms favor male applicants if past hires were mostly men.

Measurement Bias

– Data collected doesn’t accurately represent what’s being measured.
– Example: Predicting student success only from those who finish a course ignores dropouts, leading to wrong conclusions.

Stereotyping Bias

– AI perpetuates harmful stereotypes.
– Example: Language models associate “nurse” with females and “doctor” with males, reinforcing gender roles.

Out-Group Homogeneity Bias

– This bias causes an AI system to generalize individuals from underrepresented groups, treating them as more similar than they actually are.
Example: Facial recognition misidentifies people from ethnic minorities due to lack of diverse training data, risking wrongful arrests.[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Real-World Examples of Bias in AI[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Criminal Justice: COMPAS Risk Assessment Tool

Context: COMPAS is an algorithm used in the U.S. justice system to assess the likelihood of reoffending among defendants.
Bias Type: Algorithmic bias and representation bias
Issue: The algorithm systematically predicted higher recidivism risk for Black defendants compared to white defendants with similar profiles.
Consequences: Disproportionately harsher sentencing and parole decisions for marginalized communities.
Explanation: The system learned from historical crime data shaped by systemic racial profiling, which encoded past discrimination into future predictions.

Context: Generative models like DALL·E and Midjourney generate images from text prompts.
Issue: Prompts like “CEO” often generate images of white men, while prompts like “criminal” or “thug” disproportionately depict men of color.
Consequences: Reinforcement of societal stereotypes and biased public narratives.
Explanation: These models are trained on web data, where media and cultural bias are already embedded.

[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Data Privacy[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Data privacy in AI means protecting personal, identifiable, or sensitive information when data is:

Collected
Stored
Analyzed
Shared through AI systems

[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]In Monitoring, Evaluation, and Learning (MEL), data often includes:

Individual profiles
Household surveys
Community-level information

[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Privacy Risks in AI Development and Use[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Data aggregation: AI systems combine data from multiple sources, increasing the likelihood of identifying individuals from supposedly anonymized datasets.

Inference risks: Machine learning models can infer sensitive attributes such as political beliefs, sexual orientation, or health status based on other observable data (Shokri et al., 2017).

Lack of transparency: The opacity of many AI models (“black-box” nature) makes it difficult for users to understand how data is processed or how decisions are made.[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”h3″ text_direction=”default”]Regulatory and Ethical Frameworks[/nectar_responsive_text][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Globally, regulations such as the General Data Protection Regulation (GDPR) in the EU and the California Consumer Privacy Act (CCPA) provide a legal basis for protecting individuals’ data rights. These regulations emphasize data minimization, purpose limitation, and the right to explanation, all of which are relevant to AI development.

Moreover, ethical AI frameworks, including those from the OECD, UNESCO, and the IEEE, stress the need for transparency, accountability, and privacy as core design principles in AI systems.[/nectar_responsive_text][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][divider line_type=”Full Width Line” line_thickness=”1″ divider_color=”default” custom_height=”40″][/vc_column][/vc_row][vc_row type=”in_container” full_screen_row_position=”middle” column_margin=”default” column_direction=”default” column_direction_tablet=”default” column_direction_phone=”default” scene_position=”center” text_color=”dark” text_align=”left” row_border_radius=”none” row_border_radius_applies=”bg” overflow=”visible” overlay_strength=”0.3″ gradient_direction=”left_to_right” shape_divider_position=”bottom” bg_image_animation=”none”][vc_column column_padding=”no-extra-padding” column_padding_tablet=”inherit” column_padding_phone=”inherit” column_padding_position=”all” column_element_direction_desktop=”default” column_element_spacing=”default” desktop_text_alignment=”default” tablet_text_alignment=”default” phone_text_alignment=”default” background_color_opacity=”1″ background_hover_color_opacity=”1″ column_backdrop_filter=”none” column_shadow=”none” column_border_radius=”none” column_link_target=”_self” column_position=”default” gradient_direction=”left_to_right” overlay_strength=”0.3″ width=”1/1″ tablet_width_inherit=”default” animation_type=”default” bg_image_animation=”none” border_type=”simple” column_border_width=”none” column_border_style=”solid”][nectar_responsive_text inherited_font_style=”default” text_direction=”default”]Importantly, data privacy must also be considered through the lens of equity.

Data collection and AI deployment in low-income or marginalized communities may occur without adequate informed consent or clear benefit-sharing mechanisms. This risks data colonialism, where communities provide data without receiving equitable value or protection in return (Couldry & Mejias, 2019).[/nectar_responsive_text][/vc_column][/vc_row]