This contribution wants to investigate, from a theoretical-critical point of view, one of the strategically decisive faculties for the administration of the city of the future, although among the least studied: the faculty of listening. Where the metaphor of listening has been used in reference to the smart city as a place of democracy and civic participation, a technical analysis of the ways in which the city listens and eavesdrops allows us to identify points of contradiction with those representations that have been incorporated into the technological apparatus. By focusing on the case studies of acoustic monitoring devices introduced experimentally in Santander (EAR-IT) and New York (SONYC), the paper reconstructs the idea of listening underlying those systems in order to deconstruct stereotyped narrations about machine listening and artificial intelligence. In the first place, it takes in consideration the concept of “operational listening”, as a practice of machine learning that redefines human listening in its interaction with machine listening. Then it analyzes the functioning of acoustic surveillance through sound recognition, identification of the speaker and keyword spotting. Starting from those analysis, the paper individuates in the environmental and invisible features of acoustic surveillance the paradigmatic traits of a post-panoptic surveillance, that is a surveillance not oriented towards the disciplining of subjects, but to the extraction of as much data as possible to feed algorithmic prediction systems. It therefore connects the statistical methods of governance with a new securitarian configuration of the society, oriented not so much to imputation and repression, as to prediction, prevention and evaluation. This affects deeply the relation between subjectivity and privacy, since the former is ever more defined by the data it can provide, so becoming ever more a medium for machines autonomous functioning.
Questo contributo intende investigare da un punto di vista teorico-critico una delle facoltà strategicamente decisive per l’amministrazione della città del futuro, benché tra le meno studiate: la facoltà di ascoltare. Laddove la metafora dell’ascolto è stata impiegata in riferimento alla smart city come luogo della democrazia e della partecipazione civica, un’analisi tecnica dei modi in cui la città ascolta e origlia permette di individuare, incorporati negli apparati tecnologici, punti di contraddizione con quelle rappresentazioni. Partendo dai casi studio di dispositivi di monitoraggio acustico introdotti sperimentalmente a Santander (EAR-IT) e New York (SONYC), l’articolo ricostruisce l'idea di ascolto soggiacente a quei sistemi, al fine di decostruire narrazioni stereotipate sull'ascolto macchinico e sull'intelligenza artificiale. In primo luogo prende in considerazione il concetto di “ascolto operazionale” come pratica basata sul machine learning che ridefinisce l'ascolto umano nella sua interazione con l'ascolto macchinico. In secondo luogo analizza il funzionamento della sorveglianza acustica attraverso il riconoscimento sonoro, l’identificazione del parlante e l’individuazione di parole chiave. Da queste analisi l'articolo individua nella sorveglianza acustica, invisibile e ambientale, un esempio di sorveglianza post-panottica, ovvero non orientata al disciplinamento dei soggetti, ma all'estrazione di quanti più dati possibile per alimentare sistemi algoritmici di previsione. Essa mette dunque in collegamento diretto le modalità statistiche della governance con una nuova configurazione securitaria della società, orientata non tanto all'imputazione e alla repressione, quanto alla previsione, alla prevenzione e alla valutazione. Ciò incide profondamente sul rapporto tra soggettività e privacy, in quanto la prima è sempre più definita dalla sua capacità di fornire dati, diventando un medium per il funzionamento autonomo delle macchine.
Le orecchie della smart city. Riconoscimento vocale e ascolto operazionale nella "città senziente"
Domenico Napolitano
2020-01-01
Abstract
This contribution wants to investigate, from a theoretical-critical point of view, one of the strategically decisive faculties for the administration of the city of the future, although among the least studied: the faculty of listening. Where the metaphor of listening has been used in reference to the smart city as a place of democracy and civic participation, a technical analysis of the ways in which the city listens and eavesdrops allows us to identify points of contradiction with those representations that have been incorporated into the technological apparatus. By focusing on the case studies of acoustic monitoring devices introduced experimentally in Santander (EAR-IT) and New York (SONYC), the paper reconstructs the idea of listening underlying those systems in order to deconstruct stereotyped narrations about machine listening and artificial intelligence. In the first place, it takes in consideration the concept of “operational listening”, as a practice of machine learning that redefines human listening in its interaction with machine listening. Then it analyzes the functioning of acoustic surveillance through sound recognition, identification of the speaker and keyword spotting. Starting from those analysis, the paper individuates in the environmental and invisible features of acoustic surveillance the paradigmatic traits of a post-panoptic surveillance, that is a surveillance not oriented towards the disciplining of subjects, but to the extraction of as much data as possible to feed algorithmic prediction systems. It therefore connects the statistical methods of governance with a new securitarian configuration of the society, oriented not so much to imputation and repression, as to prediction, prevention and evaluation. This affects deeply the relation between subjectivity and privacy, since the former is ever more defined by the data it can provide, so becoming ever more a medium for machines autonomous functioning.I documenti in IRIS sono protetti da copyright e tutti i diritti sono riservati, salvo diversa indicazione.