Beyond the presence of AI: claim-capability discrepancies in three documented cases of AI washing
Author
Biagini, Gianmarco
Term
4. term
Publication year
2026
Submitted on
2026-10-14
Abstract
This master’s thesis investigates the phenomenon of AI washing, meaning situations where companies claim to use or possess advanced artificial intelligence capabilities that are not supported by their actual technological performance. Starting from the observation that AI is widely perceived as a strategic, transformative technology, the thesis describes how massive investments, media attention, and policy and business expectations have contributed to a strong wave of AI hype. The study identifies a research gap in understanding how discrepancies between stated AI claims and real technical capabilities emerge and manifest in practice, and formulates a research question focused on such claim–capability discrepancies in documented cases of AI washing. The theoretical framework is built around the concept of claim–capability discrepancies and a set of analytical dimensions that distinguish between what firms say about their AI and what their systems can actually do. Methodologically, the thesis employs a qualitative multiple-case study of three documented cases (Presto Automation, Delphia, and DoNotPay), drawing on documentary sources, structuring data into claim-level datasets, and analysing them abductively through within-case and cross-case analyses. As the provided excerpt does not include the results section, specific findings cannot be reported, but the thesis aims to deepen understanding of AI washing, contribute to the emerging literature on the topic, and highlight implications for researchers, practitioners, and policymakers facing growing organizational and market pressures to demonstrate AI adoption.
Denne specialeafhandling undersøger fænomenet AI-washing – når virksomheder påstår at anvende eller besidde avanceret kunstig intelligens, uden at deres faktiske teknologiske kapabiliteter lever op til disse løfter. Med udgangspunkt i, at AI i dag betragtes som en strategisk, transformerende teknologi, beskrives hvordan enorme investeringer, medieopmærksomhed og politiske samt erhvervsmæssige forventninger har skabt en stærk AI-hype. Afhandlingen identificerer et forskningsgab i forståelsen af, hvordan uoverensstemmelser mellem kommunikerede AI-påstande og reelle tekniske evner opstår og udspiller sig i praksis. For at adressere dette formuleres et forskningsspørgsmål om claim–capability discrepancies i konkrete AI-washing-tilfælde. Teoretisk tager studiet udgangspunkt i begrebet claim-capability discrepancies samt analytiske dimensioner til at skelne mellem, hvad virksomheder siger om deres AI, og hvad teknologien reelt kan. Metodisk anvendes et kvalitativt multiple-case studie af tre dokumenterede sager (Presto Automation, Delphia og DoNotPay), hvor data indsamles gennem dokumentariske kilder, struktureres i claim-niveau datasæt og analyseres ved hjælp af en abduktiv tilgang med både indenfor-case og tværgående case-analyser. Da uddraget ikke omfatter resultatafsnittet, kan de konkrete fund ikke gengives, men afhandlingen sigter mod at nuancere forståelsen af AI-washing, bidrage til den spirende litteratur på området og pege på implikationer for både forskere, praktikere og beslutningstagere i lyset af stigende organisatoriske og markedsmæssige forventninger til AI.
[This abstract has been generated with the help of AI directly from the project full text]