[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-seo-290143-105":3,"detail-sidebar-cat-0-en-105":81,"doc-detail-290143-en":131},{"code":4,"msg":5,"data":6},0,"ok",{"site_id":7,"language":8,"slug":9,"title":10,"keywords":11,"description":12,"schema_data":13,"social_meta":74,"head_meta":76,"extra_data":78,"updated_unix":80},105,"en","machine-learning-and-the-politics-of-synthetic-data","Machine learning and the politics of synthetic data","","Machine-learning algorithms are deeply embedded in contemporary society, driving scrutiny of the contents, biases, and assumptions of training datasets. The article examines what changes when training uses synthetic data—data not tied to real persons, objects, or events. Synthetic data are increasingly used across societal domains, yet their role and ethicopolitical implications remain insufficiently understood. The paper analyzes two aspects: variability exposure and shifting algorithms beyond perceived risk, showing how synthetic training reshapes machine learning’s conditions in society.",{"@graph":14,"@context":73},[15,34,56],{"@type":16,"itemListElement":17},"BreadcrumbList",[18,23,27,31],{"item":19,"name":20,"@type":21,"position":22},"https://docshare.wps.com","Home","ListItem",1,{"item":24,"name":25,"@type":21,"position":26},"https://docshare.wps.com/document/","Document",2,{"item":28,"name":29,"@type":21,"position":30},"https://docshare.wps.com/document/research-report/","Research & Report",3,{"item":32,"name":10,"@type":21,"position":33},"https://docshare.wps.com/document/machine-learning-and-the-politics-of-synthetic-data/290143/",4,{"url":32,"name":10,"@type":35,"image":36,"author":41,"headline":10,"publisher":44,"fileFormat":47,"inLanguage":8,"description":12,"dateModified":48,"datePublished":49,"encodingFormat":47,"isAccessibleForFree":50,"interactionStatistic":51},"DigitalDocument",{"url":37,"@type":38,"width":39,"height":40},"https://docshare.wps.com/thumbnails/machine-learning-and-the-politics-of-synthetic-data/290143.png","ImageObject",300,407,{"name":42,"@type":43},"Jiven","Person",{"url":19,"name":45,"@type":46},"DocShare","Organization","application/pdf","2026-10-20","2026-09-17",true,{"@type":52,"interactionType":53,"userInteractionCount":55},"InteractionCounter",{"@type":54},"ViewAction",11,{"@type":57,"mainEntity":58},"FAQPage",[59,65,69],{"name":60,"@type":61,"acceptedAnswer":62},"What problem does the article address about synthetic data in machine learning?","Question",{"text":63,"@type":64},"It asks what happens when machine-learning algorithms are trained on data that are synthetic rather than derived from real persons, objects, or events, and what ethicopolitical implications follow.","Answer",{"name":66,"@type":61,"acceptedAnswer":67},"What two central aspects does the paper use to explore the politics of synthetic data?",{"text":68,"@type":64},"It focuses on how synthetic data promise to provide rich exposure to variability for algorithms, and how synthetic data promise to place algorithms beyond the realm of risk.",{"name":70,"@type":61,"acceptedAnswer":71},"How is ConfigNet used to generate synthetic training data in the paper’s example?",{"text":72,"@type":64},"The article describes ConfigNet as a model that generates photorealistic synthetic human face images, enabling scalable generation and enhanced control over features such as pose, hair, expression, and illumination for facial recognition training.","https://schema.org",{"og:url":32,"og:type":75,"og:title":10,"og:site_name":45,"og:description":12},"article",{"robots":77,"canonical":32},"index,follow",{"doc_id":79,"site_id":7},290143,1789640792,{"code":4,"msg":82,"data":83},"success",[84,88,92,96,101,106,111,115,120,123,127],{"id":22,"doc_module":4,"doc_module_name":25,"category_name":85,"show_sort_weight":86,"slug":87},"Story & Novel",90,"story-novel",{"id":26,"doc_module":4,"doc_module_name":25,"category_name":89,"show_sort_weight":90,"slug":91},"Literature",80,"literature",{"id":33,"doc_module":4,"doc_module_name":25,"category_name":93,"show_sort_weight":94,"slug":95},"Exam",70,"exam",{"id":97,"doc_module":4,"doc_module_name":25,"category_name":98,"show_sort_weight":99,"slug":100},5,"Comic",60,"comic",{"id":102,"doc_module":4,"doc_module_name":25,"category_name":103,"show_sort_weight":104,"slug":105},6,"Technology",50,"technology",{"id":107,"doc_module":4,"doc_module_name":25,"category_name":108,"show_sort_weight":109,"slug":110},7,"Healthcare",40,"healthcare",{"id":112,"doc_module":4,"doc_module_name":25,"category_name":29,"show_sort_weight":113,"slug":114},8,30,"research-report",{"id":116,"doc_module":4,"doc_module_name":25,"category_name":117,"show_sort_weight":118,"slug":119},9,"Religion & Spirituality",20,"religion-spirituality",{"id":118,"doc_module":4,"doc_module_name":25,"category_name":121,"show_sort_weight":118,"slug":122},"World Cup","world-cup",{"id":124,"doc_module":4,"doc_module_name":25,"category_name":125,"show_sort_weight":124,"slug":126},10,"Lifestyle","lifestyle",{"id":128,"doc_module":4,"doc_module_name":25,"category_name":129,"show_sort_weight":97,"slug":130},19,"General","general",{"code":4,"msg":82,"data":132},{"doc_id":79,"user_id":133,"nickname":42,"user_avatar":134,"doc_module":4,"category_id":112,"category_name":29,"doc_title":10,"doc_description":12,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":55,"is_deleted":4,"is_public":22,"is_downloadable":22,"audit_status":22,"page_count":140,"language":141,"language_code":8,"site_id":7,"html_lang":8,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":12,"update_tm":80,"read_time":113,"preview_page_count":4},1099513958607,"https://ap-avatar.wpscdn.com/avatar/100002390cf8733938c?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778829742770036399","Original Research Article  \nMachine learning and the politics of synthetic data  \nBenjamin N Jacobsen   \nBig Data & Society January–June: 1–12  \n© The Author(s) 2023  \nArticle reuse guidelines: [sagepub.com/journals-permissions](sagepub.com/journals-permissions)[ ](sagepub.com/journals-permissions)[DOI: 10.1177/20539517221145372](DOI: 10.1177/20539517221145372)[ ](DOI: 10.1177/20539517221145372)[journals.sagepub.com/home/bds](journals.sagepub.com/home/bds)  \nAbstract  \nMachine-learning algorithms have become deeply embedded in contemporary society. As such, ample attention has been paid to the contents, biases, and underlying assumptions of the training datasets that many algorithmic models are trained on. Yet, what happens when algorithms are trained on data that are not real, but instead data that are ‘synthetic’, not referring to real persons, objects, or events? Increasingly, synthetic data are being incorporated into the training of machine-learning algorithms for use in various societal domains. There is currently little understanding, however, of the role played by and the ethicopolitical implications of synthetic training data for machine-learning algorithms. In this article, I explore the politics of synthetic data through two central aspects: first, synthetic data promise to emerge asa rich source of exposure to variability for the algorithm. Second, the paper explores how synthetic data promise to place algorithms beyond the realm of risk. I propose that an analysis of these two areas will help us better understand the ways in which machine-learning algorithms are envisioned in the light of synthetic data, but also how synthetic training data actively reconfigure the conditions of possibility for machine learning in contemporary society.  \nKeywords  \nMachine learning, data, algorithms, risk, ethics, variability  \nIntroduction  \nIn 2020, computer scientists at the Mixed Reality & AI Labs at Microsoft Cambridge developed an algorithmic model that is capable of generating photorealistic synthetic images of human faces. The aim of the model is to provide a means of developing synthetic training data for facial recognition algorithms. The model, which is called ConfigNet, is a composite of computer graphics techniques and generative adversarial networks (GANs), the algorithm often evoked in the context of so-called ‘deepfakes’ (de Vries, 2020) . The key idea behind ConfigNet, the researchers claim, is to enable both the generation of photorealistic synthetic faces at scale as well as enhanced control over the generative process: ‘this allows for independent control of various face aspects including: head pose, hair style, facial hair style, expression and illumination’ (Kowalski et al., 2020: 301). ‘By simultaneously training the model on unlabelled face images’, the computer scientists explain, ‘it learns to generate photorealistic looking faces, while enabling full control over these outputs’ (p. 301). As a result, with ConfigNet the synthetic face is rendered into an array of controllable and tweakable parameters within the algorithmic model: head poses can be shifted, hair styles can be changed, skin colours can be altered. A question raised by  \nMicrosoft Cambridge’s ConfigNet model is therefore: What if facial recognition algorithms could be trained using so-called ‘synthetic faces’, faces that do not relate to any specific persons? Synthetic data in this context promise to provide machine-learning engineers both with control over the manufacturing process as well as with access to a vast heterogeneity of faces that can be used to train facial recognition systems – none of which refer to any actual real persons. In short, whatever face is needed can be algorithmically generated. The parameters can be tweaked to give machine-learning models any types of faces they need.  \nThe importance of data for the reconfiguration of society has been well documented in the critical scholarship (Beer, 2016; Boyd and Crawford, 2012; Halpe","cbCaimC4WgkptoDI","https://ap.wps.com/l/cbCaimC4WgkptoDI","pdf",595853,12,"English","# Abstract\n# Introduction","[{\"question\":\"What problem does the article address about synthetic data in machine learning?\",\"answer\":\"It asks what happens when machine-learning algorithms are trained on data that are synthetic rather than derived from real persons, objects, or events, and what ethicopolitical implications follow.\"},{\"question\":\"What two central aspects does the paper use to explore the politics of synthetic data?\",\"answer\":\"It focuses on how synthetic data promise to provide rich exposure to variability for algorithms, and how synthetic data promise to place algorithms beyond the realm of risk.\"},{\"question\":\"How is ConfigNet used to generate synthetic training data in the paper’s example?\",\"answer\":\"The article describes ConfigNet as a model that generates photorealistic synthetic human face images, enabling scalable generation and enhanced control over features such as pose, hair, expression, and illumination for facial recognition training.\"}]","Machine learning and the politics of synthetic data | PDF"]