In this paper, we disentangle the changes that the rise of artificial intelligence (AI) technologies is inducing in the semiconductor industry. Chips based on the von Neumann architecture are struggling to deliver performance across a wide range of applications, and the new AI segment is only adding to this struggle. This poses a new challenge to chip design, with flexibility of computation at its core, i.e., hardware’s ability to support a large software variety, rather than computation speed. We identify and analyze forces and mechanisms at work and discuss the product configurations which could characterize the future of the semiconductor industry. We outline two possible scenarios: (i) fragmentation of the semiconductor industry into submarkets with dedicated chips and (ii) the shift of the industry to a system-on-a-chip-based dominant design with the emergence of a new platform chip. We rationalize the unfolding situation by modeling consumer choice between computing systems based on their crucial characteristics—speed, flexibility, and energy efficiency.
In this paper, we offer an original framework to study Artificial Intelligence (AI). The perspective we propose is based on the idea that AI is a system technology, and that a useful description of AI cannot abstain from mapping the components of the system, their interdependence, and how the synergies they create shape at the roots the directions of AI development. We adopt the concept of Large Technical System (LTS) to give substance and structure to our idea. Using LTS, we are able to scaffold AI and the forces at work steering its production, deployment, and evolution. We find that AI as a system shares essential features with infrastructural technologies such as the Internet. The LTS framework proves very useful to capture important nuances of the technology, and it allows us to trace the connections and cross-influences among its constituting domains - algorithms (software), compute (hardware), and data. We compare our proposed framework with other concepts usually associated with radical innovations, and suggest in which respects AI differs from these ideal-types. We consider ours a timely exercise, as we witness the formation of an AI industry. While in the making, this industry is rapidly ossifying, together with its specific problems, power imbalances, and development scenarios; the focus on the system-ness of AI allows uncovering the deeper structure of this technological breakthrough.
scite is a Brooklyn-based organization that helps researchers better discover and understand research articles through Smart Citations–citations that display the context of the citation and describe whether the article provides supporting or contrasting evidence. scite is used by students and researchers from around the world and is funded in part by the National Science Foundation and the National Institute on Drug Abuse of the National Institutes of Health.