Join us for a guest seminar by Dr Vassilis Galanos (University of Stirling) on the fascinating history of AI and Hype. This hybrid event will be held at the Bristol Digital Futures Institute and online. (9^th July 12-2)

Event schedule
12- 1: webinar and Q&A (hybrid)
1-2: workshop activity over lunch (in person only)

Please register here to help us with planning of catering: https://www.eventbrite.com/e/guest-seminar-by-vassilis-galanos-ai-hype-studies-tickets-1380867173889?aff=oddtdtcreator

 

Abstract: To Have Done with The Metaphor of Summers and Winters: Can AI And Internet History Cure Hype?

This paper addresses the often-misleading discourse around artificial intelligence (AI), arguing that the intertwined evolution of AI and internet technologies requires more explicit attention to avoid what Rayner terms “novelty traps” - perceived innovation grounded more in language and hype than substantial technical breakthroughs. Building on, yet critiquing, Grudin’s historical heuristic of alternating AI winters and human-computer interaction (HCI) summers, I demonstrate that this model neglects the concurrent, co-dependent developments in both AI and internet domains, as well as the socio-linguistic motives behind naming and categorisation - issues central to Latour's five tactics of reshuffling goals and interests and Olazaran’s distinction between “official” and “research area” technological histories. Empirically, I analyse three pivotal historical shifts: (1) the rise of digital object architecture from packet switching origins, shaping vital data structures for neural networks and exposing divergent AI and internet developmental paths, especially through cases like the perceptron debate and research centres such as Edinburgh, ARPANET, and Toronto’s neural network group; (2) the emergence of ImageNet, tracing its semantic lineage from WordNet, charting its role in transitioning from Web 2.0 rhetoric to the contemporary framing of AI as mainly machine learning, sidelining symbolic AI; (3) the recent deployment of large language models as commercial search interfaces, which illustrates the inseparability of AI and internet platforms today. Complementing this analysis, I review overlapping discourses in computer, information, and AI ethics to emphasise underlying shared technical and ethical dimensions. I conclude by proposing to portmanteau terms to analyse hype: (1) hybernetics: the view of technology's history as led by multiple conjoined or convoluted narratives, filled with reciprocal influences and complex hype feedback systems; and (2) eggspoontations: the synthesis of extremely fast and extremely slow experience of digital technological duration as experienced by highly hype-mediated environments.