--- title: Global Bias in AI: When Western Data Shapes the World -... description: A conference talk exploring how Western-centric training data creates systemic bias in AI systems across image recognition, medical AI, and financial... canonical: https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw twitter:card: summary_large_image twitter:site: @dailydotdev og:type: website og:site_name: daily.dev og:title: Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026 | daily.dev og:description: A conference talk exploring how Western-centric training data creates systemic bias in AI systems across image recognition, medical AI, and financial... og:url: https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw og:image: https://api.daily.dev/og/posts/28Z5YSNMw.png og:image:alt: Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026 og:image:width: 1200 og:image:height: 630 og:locale: en --- > ## Documentation Index > Fetch the complete documentation index at: https://daily.dev/llms.txt > Use this file to discover all available pages before exploring further. # Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026 **[NDC Conferences](https://daily.dev/sources/ndc)** · 56 min read · 0 upvotes · 0 comments ## Summary A conference talk exploring how Western-centric training data creates systemic bias in AI systems across image recognition, medical AI, and financial algorithms. Key case studies include Google Photos misclassifying Black faces as gorillas (with the label still silenced 8 years later), Stable Diffusion amplifying racial and gender stereotypes in job-role image generation, pulse oximeters missing dangerously low oxygen levels in Black patients at nearly 3x the rate, the Fitzpatrick skin scale excluding darker skin tones until 1988, Western credit-scoring models failing economically active people in East Africa who use mobile money systems, and government algorithms in the Netherlands, Australia, and India causing serious harm through encoded bias. The talk also covers data annotator exploitation (Kenyan workers paid under $2/hour for traumatic content labeling), the WEIRD problem in psychology applied to AI, model collapse, and the absence of African nations from OECD AI governance. African fintech companies like M-KOPA, Tala, and Jumo are highlighted as examples of locally-appropriate alternatives. ## Full article daily.dev links to this article rather than hosting it. Read it at the original source: ## Questions this post answers ### What did the Bloomberg study on Stable Diffusion find about racial and gender bias in AI-generated images? Bloomberg generated over 5,000 images across 14 job titles (300 per role) using Stable Diffusion and found high-paying professions like doctors, lawyers, and engineers were skewed toward lighter-skinned men, while low-paying roles like janitor and fast food worker skewed toward women and darker-skinned people. For crime keywords, over 80% of inmate images and more than half of drug dealer images featured darker skin tones, and terrorist prompts defaulted to men with beards and head coverings. _Developers building or evaluating image generation pipelines track findings like these on daily.dev._ ### What was wrong with the Fitzpatrick skin tone scale used in medical AI datasets? The Fitzpatrick scale, created by a Harvard dermatologist in 1975, originally had only four categories covering lighter skin tones. The two darkest skin types (types 5 and 6) were not added until 1988, 13 years later. Despite this gap, it became the default skin classification system in medical AI datasets. The Monk Skin Tone Scale, released in 2022 with 10 tones, is a more representative alternative now being promoted by Google. _Medical AI and health-tech developers keeping up with bias research and dataset standards find relevant coverage on daily.dev._ ### Why do Western credit scoring AI models fail for people in East Africa? Western credit scoring models look for bank accounts, credit card history, utility bills, and tax filings. In East Africa, most transactions run through mobile money systems like M-Pesa, and assets such as land and gold are often communally shared. These models see no recognizable financial signals and flag economically active, reliable people as uncreditworthy. African fintech companies like M-KOPA, Tala, and Jumo have addressed this using mobile and behavioral data instead. _Engineers building fintech or credit systems for emerging markets find relevant case studies and alternatives on daily.dev._ ## Similar posts on daily.dev - [Bias & the Myth of the Objective Machine](https://daily.dev/posts/bias-the-myth-of-the-objective-machine-xgbnfmvgd) · Atomic Spin · 0 upvotes · 0 comments - [Did we manage to solve the gender bias in AI yet?](https://daily.dev/posts/did-we-manage-to-solve-the-gender-bias-in-ai-yet--05dp6y3ye) · Code Like A Girl · 0 upvotes · 0 comments - [TCS\+ \| Flipping the narrative on AI in the Global South](https://daily.dev/posts/tcs-flipping-the-narrative-on-ai-in-the-global-south-yzimz0mf2) · TechCentral · 0 upvotes · 0 comments - [Most users cannot identify AI bias, even in training data](https://daily.dev/posts/most-users-cannot-identify-ai-bias-even-in-training-data-tzmenkyaq) · Hacker News · 0 upvotes · 0 comments --- Tags: [#ai-governance](https://daily.dev/tags/ai-governance), [#ethical-ai](https://daily.dev/tags/ethical-ai), [#machine-learning](https://daily.dev/tags/machine-learning) [View this post on daily.dev](https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw) ```json {"@context":"https://schema.org","@graph":[{"@type":"Organization","@id":"https://daily.dev/#organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180},"sameAs":["https://twitter.com/dailydotdev","https://github.com/dailydotdev","https://www.linkedin.com/company/daily-dev-ltd"]},{"@type":"WebSite","@id":"https://daily.dev/#website","url":"https://daily.dev","name":"daily.dev","publisher":{"@id":"https://daily.dev/#organization"},"potentialAction":{"@type":"SearchAction","target":{"@type":"EntryPoint","urlTemplate":"https://daily.dev/search?q={search_term_string}"},"query-input":"required name=search_term_string"}}]} {"@context":"https://schema.org","@type":"TechArticle","headline":"Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026","url":"https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw","mainEntityOfPage":{"@type":"WebPage","@id":"https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw"},"datePublished":"2026-08-11T08:27:52.805Z","dateModified":"2026-10-14T02:28:52.604Z","description":"A conference talk exploring how Western-centric training data creates systemic bias in AI systems across image recognition, medical AI, and financial...","image":"https://i.ytimg.com/vi/FCgP5xLsHgQ/sddefault.jpg","thumbnailUrl":"https://i.ytimg.com/vi/FCgP5xLsHgQ/sddefault.jpg","isAccessibleForFree":true,"articleSection":"NDC Conferences","inLanguage":"en","publisher":{"@type":"Organization","name":"daily.dev","url":"https://daily.dev","logo":{"@type":"ImageObject","url":"https://daily.dev/apple-touch-icon.png","width":180,"height":180}},"author":{"@type":"Organization","name":"NDC Conferences","logo":"https://media.daily.dev/image/upload/s--SiIyojEj--/f_auto/v1748760369/logos/ndc","url":"https://daily.dev/sources/ndc"},"commentCount":0,"discussionUrl":"https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw","interactionStatistic":[{"@type":"InteractionCounter","interactionType":{"@type":"LikeAction"},"userInteractionCount":0},{"@type":"InteractionCounter","interactionType":{"@type":"CommentAction"},"userInteractionCount":0}],"keywords":"ai-governance,ethical-ai,machine-learning","timeRequired":"PT56M","video":{"@type":"VideoObject","name":"Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026","description":"A conference talk exploring how Western-centric training data creates systemic bias in AI systems across image recognition, medical AI, and financial...","thumbnailUrl":"https://i.ytimg.com/vi/FCgP5xLsHgQ/sddefault.jpg","uploadDate":"2026-08-11T08:27:52.805Z","duration":"PT56M","url":"https://api.daily.dev/r/28Z5YSNMw","embedUrl":"https://www.youtube.com/embed/FCgP5xLsHgQ"}} {"@context":"https://schema.org","@type":"BreadcrumbList","itemListElement":[{"@type":"ListItem","position":1,"name":"Home","item":"https://daily.dev"},{"@type":"ListItem","position":2,"name":"NDC Conferences","item":"https://daily.dev/sources/ndc"},{"@type":"ListItem","position":3,"name":"Global Bias in AI: When Western Data Shapes the World - Masuma Shariff - NDC Copenhagen 2026"}]} {"@context":"https://schema.org","@type":"FAQPage","@id":"https://daily.dev/posts/global-bias-in-ai-when-western-data-shapes-the-world---masuma-shariff---ndc-copenhagen-2026-28z5ysnmw#faq","mainEntity":[{"@type":"Question","name":"What did the Bloomberg study on Stable Diffusion find about racial and gender bias in AI-generated images?","acceptedAnswer":{"@type":"Answer","text":"Bloomberg generated over 5,000 images across 14 job titles (300 per role) using Stable Diffusion and found high-paying professions like doctors, lawyers, and engineers were skewed toward lighter-skinned men, while low-paying roles like janitor and fast food worker skewed toward women and darker-skinned people. For crime keywords, over 80% of inmate images and more than half of drug dealer images featured darker skin tones, and terrorist prompts defaulted to men with beards and head coverings. Developers building or evaluating image generation pipelines track findings like these on daily.dev."}},{"@type":"Question","name":"What was wrong with the Fitzpatrick skin tone scale used in medical AI datasets?","acceptedAnswer":{"@type":"Answer","text":"The Fitzpatrick scale, created by a Harvard dermatologist in 1975, originally had only four categories covering lighter skin tones. The two darkest skin types (types 5 and 6) were not added until 1988, 13 years later. Despite this gap, it became the default skin classification system in medical AI datasets. The Monk Skin Tone Scale, released in 2022 with 10 tones, is a more representative alternative now being promoted by Google. Medical AI and health-tech developers keeping up with bias research and dataset standards find relevant coverage on daily.dev."}},{"@type":"Question","name":"Why do Western credit scoring AI models fail for people in East Africa?","acceptedAnswer":{"@type":"Answer","text":"Western credit scoring models look for bank accounts, credit card history, utility bills, and tax filings. In East Africa, most transactions run through mobile money systems like M-Pesa, and assets such as land and gold are often communally shared. These models see no recognizable financial signals and flag economically active, reliable people as uncreditworthy. African fintech companies like M-KOPA, Tala, and Jumo have addressed this using mobile and behavioral data instead. Engineers building fintech or credit systems for emerging markets find relevant case studies and alternatives on daily.dev."}}]} ```