The dissertation investigates the connections between artificial intelligence and contemporary art, mapping the relationships between creative practices and digital art historical repositories, and examining how artists and cultural institutions integrate neural networks and art-historical datasets into creative processes. The work is structured around four lines of inquiry: a historical reconstruction of the development of neural networks and their impact on art, with particular attention to experiments in programmed art and computer art; an analysis of the role of public data and open digital collections in training AI models, with consideration of ethical implications; a study of knowledge representation within the “latent space” of neural networks, including the definition of an evaluation dataset for multimodal models (ArtVision) consisting of 250 tasks in eight categories based on art history; and finally, a critical register of 170 artists who, between 2014 and 2024, have employed neural networks, organized into generational and geographical clusters to reflect on transformations in contemporary artistic production. The research emphasizes how developments in artificial intelligence and neural networks have influenced artistic experimentation, highlighting how such systems reflect the symbolic structures embedded in training choices, and how artistic practices can serve as a means to interpret and understand emerging technologies and their cultural applications.
La tesi indaga le connessioni tra intelligenza artificiale e arte contemporanea, tracciando una mappa dei rapporti tra pratiche creative e repertori digitali di storia dell’arte, analizzando come artisti e istituzioni culturali integrino reti neurali e dataset storico-artistici nei processi creativi. Il lavoro si articola in quattro linee di ricerca: una ricostruzione storica dell’evoluzione delle reti neurali e del loro impatto sull’arte, con particolare attenzione alle sperimentazioni di arte programmata e computer art; un’analisi del ruolo dei dati pubblici e delle collezioni digitali aperte nell’addestramento dei modelli di IA, con attenzione alle ricadute etiche; uno studio sulla rappresentazione della conoscenza nello "spazio latente" delle reti neurali che include la definizione di un dataset di valutazione per modelli multimodali (ArtVision) composto da 250 task in otto categorie basato sulla storia dell'arte; infine, un regesto critico di 170 artisti che, tra il 2014 e il 2024, hanno utilizzato reti neurali, organizzati in cluster generazionali e geografici per riflettere sulle trasformazioni della produzione artistica contemporanea. La ricerca sottolinea come la ricerca sull'intelligenza artificiale e le reti neurali abbiano influenzato le sperimentazioni artistiche, evidenziando come riflettano le strutture simboliche presenti nelle scelte in fase di addestramento e come le pratiche artistiche possano essere un elemento per interpretare e comprendere le tecnologie emergenti e le loro applicazioni culturali.
Le intelligenze artificiali nel panorama artistico contemporaneo. Potenzialità e criticità dell’uso delle reti neurali e di grandi dataset storico-artistici
DE GASPERIS, PAOLO
2026
Abstract
The dissertation investigates the connections between artificial intelligence and contemporary art, mapping the relationships between creative practices and digital art historical repositories, and examining how artists and cultural institutions integrate neural networks and art-historical datasets into creative processes. The work is structured around four lines of inquiry: a historical reconstruction of the development of neural networks and their impact on art, with particular attention to experiments in programmed art and computer art; an analysis of the role of public data and open digital collections in training AI models, with consideration of ethical implications; a study of knowledge representation within the “latent space” of neural networks, including the definition of an evaluation dataset for multimodal models (ArtVision) consisting of 250 tasks in eight categories based on art history; and finally, a critical register of 170 artists who, between 2014 and 2024, have employed neural networks, organized into generational and geographical clusters to reflect on transformations in contemporary artistic production. The research emphasizes how developments in artificial intelligence and neural networks have influenced artistic experimentation, highlighting how such systems reflect the symbolic structures embedded in training choices, and how artistic practices can serve as a means to interpret and understand emerging technologies and their cultural applications.| File | Dimensione | Formato | |
|---|---|---|---|
|
Tesi_dottorato_DeGasperis.pdf
accesso aperto
Licenza:
Creative Commons
Dimensione
10.87 MB
Formato
Adobe PDF
|
10.87 MB | Adobe PDF | Visualizza/Apri |
I documenti in UNITESI sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.
https://hdl.handle.net/20.500.14242/357533
URN:NBN:IT:UNIROMA1-357533