BEGIN:VCALENDAR VERSION:2.0 PRODID:-//wp-events-plugin.com//7.4.5//EN TZID:Europe/Paris X-WR-TIMEZONE:Europe/Paris BEGIN:VEVENT UID:0-315@lisn.upsaclay.fr DTSTART;TZID=Europe/Paris:20251002T140000 DTEND;TZID=Europe/Paris:20251002T140000 DTSTAMP:20250925T145713Z URL:https://www.lisn.upsaclay.fr/evenements/construction-of-model-behavior -by-novices-through-interaction-with-machine-learning/ SUMMARY:Construction of Model Behavior by Novices through Interaction with Machine Learning DESCRIPTION:Jury\nGenoveva Vargas-Solar – Researcher (HDR)\, Université Claude Bernard Lyon 1 – Reviewer &\; Examiner\nSimone Stumpf –  \;Professor\, University of Glasgow – Reviewer &\; Examiner\nMarianel a Ciolfi Felice – \;Assistant Professor\, KTH Royal Institute of Tec hnology – Examiner\nAnastasia Bezerianos – Professor\, Université Par is-Saclay – Examiner\nAbstract\nMachine learning (ML) systems increasing ly mediate decisions across domains such as hiring\, healthcare\, criminal justice\, and content moderation. While often framed as neutral or object ive\, these systems frequently reproduce the very inequalities they claim to address. Such harms are not isolated technical failures\, but reflectio ns of broader sociotechnical dynamics: models encode dominant norms shaped by the data\, assumptions\, and institutional practices that underlie the ir design. Addressing these dynamics requires moving beyond internal perfo rmance metrics and attending to how model behavior unfolds in context\, in cluding how models operate in real-world settings and how people interpret and respond to their outputs.\nThis thesis investigates how model behavio r is not only examined but constructed through human interaction with ML m odels in real-world settings. It centers novice users who are typically ex cluded from formal development pipelines but directly impacted by model de cisions\, exploring how they interpret\, evaluate\, and potentially shape model behavior through interaction. Grounded in Human–Computer Interacti on (HCI) and informed by Science and Technology Studies (STS) and AI ethic s\, this thesis adopts a human-centered lens to present two empirical stud ies that engage novice users in the development and post-deployment phases of the ML lifecycle. These studies examine how user interaction makes mod el behavior accessible for reflection\, negotiation\, critique\, and\, if possible\, change.\nThe first study introduces Collaborative Interactive M achine Teaching \;(CIMT)\, where small groups of novice users train an image classifier using a web-based platform called TeachTOK. It examines how participants collaboratively plan teaching strategies\, curate trainin g data\, and reflect on model behavior using real-time feedback. Findings show that collaboration was central to negotiating teaching objectives\, s uch as representational diversity\, and building a shared understanding of model behavior.\nThe second study investigates user-driven algorithm audi ting \;as a post-deployment practice\, in which non-experts evaluate a pretrained image captioning model for gender bias. Drawing on cognitive m odels of sensemaking\, we designed an auditing interface with two addition al tools: one that supports hypothesis testing through input manipulation\ , and another that enables grouping outputs to reveal patterns. A between- subjects online study examined how interface conditions shaped users’ se nsemaking outcomes. Findings show that different tools influenced which ha rms users could detect\, how they formed hypotheses\, and how they develop ed confidence in their interpretations. \;\nAcross both development an d post-deployment phases\, this thesis demonstrates how novices’ interac tion with the ML model contributes to the construction of model behavior. In CIMT\, participants shaped behavior through teaching and negotiation. I n auditing\, they surfaced harms and evaluated model behavior through iter ative sensemaking. In both cases\, users brought their own perspectives to bear on how models should behave\, perspectives often excluded from conve ntional ML development and evaluation.\nThis work contributes empirical\, technical\, and methodological insights that have the potential to support participatory and reflexive approaches to machine learning within human-c entered AI\, and calls for future research grounded in sustained\, real-wo rld\, and community-led practices. CATEGORIES:IaH,Thèses et HDR LOCATION:LISN Site Belvédère\, Rue du Belvédère Campus universitaire b ât 507 91405 Orsay\, France X-APPLE-STRUCTURED-LOCATION;VALUE=URI;X-ADDRESS=Rue du Belvédère Campus universitaire bât 507 91405 Orsay\, France;X-APPLE-RADIUS=100;X-TITLE=LIS N Site Belvédère:geo:0,0 END:VEVENT BEGIN:VTIMEZONE TZID:Europe/Paris X-LIC-LOCATION:Europe/Paris BEGIN:DAYLIGHT DTSTART:20250330T030000 TZOFFSETFROM:+0100 TZOFFSETTO:+0200 TZNAME:CEST END:DAYLIGHT END:VTIMEZONE END:VCALENDAR