Challenges competitions and prizes have long been used to advance or accelerate technological innovation. The world of robotics and automation is no different: the DARPAGrand Challenge first hosted in 2004 kick-started the development of autonomous vehicles. Yet as AI research itself has proliferated moving away from state-funded programmes to all manner of start-up research centre and big tech-driven initiatives the practice of hosting challenges has only intensified.

This workshop examines the phenomenon of such challenges as they seek to stimulate and structure AI autonomous vehicle and machine vision research in the field. The workshop considers the value of challenges to scientific and technological problem-solving and to the pursuit of scientific and technological solutions. It also aims to explore the value of challenges both to challenge organizers and challenge participants: why host them? Why participate? In this the workshop seeks to examine the political economy of AI research and how the terms of participation in such challenges are carefully prescribed by organizers. Challenges not only alert hosts to emerging talent but help in establishing feeder networks often constituting a successful ‘conveyor belt’ of computer scientists machine vision researchers and software engineers for organizers. Likewise in distributing and externalizing specific AI tasks challenges shrink associated labour costs of performing AI work to nominal levels. In participating however ordinarily young or novice researchers are offered the opportunity to tackle ‘cutting-edge’ industry problems with the promise of attracting the attention of a big tech company at the end.

Objective

• Examine machine vision ‘challenges’ in autonomous vehicle research
• Advance contemporary work on the political economy of AI in respect to how datamachine-learning cloud infrastructures computation highly-skilled labour start-ups and big tech firms contribute to the development of AI.