Neural Machine Translation: From Commodity to Commons?
Résumé
When free and instant translation is available to many, this may paradoxically render translation ubiquitous and obsolete. Custom Neural Machine Translation (NMT) tools are available for those who need higher quality and more secure translation. The complex algorithms and voluminous training data mean that only larger language-service providers can meet specific standards, which increases oligopolistic market trends. There are a number of issues with NMT including the lack of accountability and increasing standardization/erasure of language(s), propagation of fake content, censorship. Recasting machine translation into an ecosystem of digital authority (Vitali-Rosati, On editorialization, structuring space and authority in the digital age, Institute of Network Cultures, 2018) and building on knowledge as a commons (Hesse and Ostrom, Understanding knowledge as a commons: From theory to practice, MIT Press, 2007), we may conceive of translation as a public utility rather than a commodity (Enriquez-Raido, IJoC, 10, 970–988, 2016).