Artificial Intelligence is transforming production, decision-making, and creative processes in many sectors. But what about its environmental and social costs? This thesis critically surveys the concept of 'AI pollution,' much of which is divided into three main areas. The first is logical pollution, which concerns bias, distortion, and the spread of misinformation due to incomplete or biased datasets and black-box algorithmic models. The second is physical pollution, caused by the intense consumption of energy at data centers, the high water intake for cooling processes, and carbon emissions through non-renewable energy sources. The third is material pollution, which refers to environmental damages and health hazards that accompany the manufacturing, replacement, and disposal of electronic hardware used in AI. The research further touches upon the psychological, professional, and cultural consequences of mass AI adoption, concerning human creativity, cognitive substitution in danger, and the transformation of work and identity. The thesis concludes with an evaluation of key mitigation strategies, including sustainable server infrastructure, renewable energy adoption, circular hardware design, and ethical regulatory frameworks. It ultimately reflects on the responsibility of companies, developers, and institutions in shaping a more sustainable, transparent, and accountable technological business model.
L’intelligenza artificiale sta trasformando la produzione, i processi decisionali e quelli creativi in numerosi settori. Ma quali sono i suoi costi ambientali e sociali? Questa tesi analizza in modo critico il concetto di “inquinamento dell’IA”, suddiviso in tre principali aree tematiche. La prima è l’inquinamento logico, che riguarda i bias, le distorsioni e la diffusione di disinformazione derivanti da dataset incompleti o parziali e da modelli algoritmici opachi (black-box). La seconda è l’inquinamento fisico, causato dall’intenso consumo di energia dei data center, dal massiccio utilizzo di acqua per i processi di raffreddamento e dalle emissioni di carbonio associate all’uso di fonti energetiche non rinnovabili. La terza è l’inquinamento materiale, che si riferisce ai danni ambientali e ai rischi per la salute connessi alla produzione, sostituzione e smaltimento dell’hardware elettronico utilizzato nell’IA. La ricerca affronta anche le conseguenze psicologiche, professionali e culturali derivanti dalla diffusione massiva dell’IA, come l’indebolimento della creatività umana, il rischio di sostituzione cognitiva e la trasformazione del lavoro e dell’identità. La tesi si conclude con una valutazione delle principali strategie di mitigazione, tra cui infrastrutture server sostenibili, adozione di energie rinnovabili, progettazione circolare dell’hardware e quadri normativi etici. In ultima analisi, si riflette sulla responsabilità di imprese, sviluppatori e istituzioni nel promuovere un modello di business tecnologico più sostenibile, trasparente e responsabile.
Smart but Dirty: The Hidden Environmental and Cognitive Costs of Algorithmic Intelligence
FESTOSI, MARIO
2024/2025
Abstract
Artificial Intelligence is transforming production, decision-making, and creative processes in many sectors. But what about its environmental and social costs? This thesis critically surveys the concept of 'AI pollution,' much of which is divided into three main areas. The first is logical pollution, which concerns bias, distortion, and the spread of misinformation due to incomplete or biased datasets and black-box algorithmic models. The second is physical pollution, caused by the intense consumption of energy at data centers, the high water intake for cooling processes, and carbon emissions through non-renewable energy sources. The third is material pollution, which refers to environmental damages and health hazards that accompany the manufacturing, replacement, and disposal of electronic hardware used in AI. The research further touches upon the psychological, professional, and cultural consequences of mass AI adoption, concerning human creativity, cognitive substitution in danger, and the transformation of work and identity. The thesis concludes with an evaluation of key mitigation strategies, including sustainable server infrastructure, renewable energy adoption, circular hardware design, and ethical regulatory frameworks. It ultimately reflects on the responsibility of companies, developers, and institutions in shaping a more sustainable, transparent, and accountable technological business model.| File | Dimensione | Formato | |
|---|---|---|---|
|
Festosi Mario.pdf
Accesso riservato
Dimensione
11.06 MB
Formato
Adobe PDF
|
11.06 MB | Adobe PDF |
The text of this website © Università degli studi di Padova. Full Text are published under a non-exclusive license. Metadata are under a CC0 License
https://hdl.handle.net/20.500.12608/94685