How Machine Vision Challenges Structure AI Innovation

Challenges, competitions and prizes have long played a role in driving technological innovation. In the late 1980s, the concept of ‘Grand Challenges’ emerged as a framework for realising research in science and technology. More precisely, Raj Reddy’s 1988 Presidential Address to the Association for the Advancement of Artificial Intelligence (AAAI) aimed to propel AI research towards tangible, concrete applications. Chess-playing machines and autonomous vehicles were presented, amongst others, as ‘bold national initiatives’ intended to ‘capture the imagination of the public’ (Reddy, 1988: 18). Despite 25 years of financial support from the likes of the Defense Advanced Research Projects Agency (DARPA), National Science Foundation (NSF), and NASA, Reddy contended that AI research—in the US, at least—now needed to enter an ‘era of accountability’ (Reddy, 1988: 9).

In recent years, talk of the ‘Grand Challenges’ of AI has receded. In its place—as the era of accountability turns into an era of accumulation—a series of altogether less grand challenges. These might, instead, be understood as ‘incremental challenges’: a host of competitions organised by start-ups, research centres, and platform firms to facilitate cutting-edge innovation in AI, machine learning (ML) and machine vision more specifically (Hind et al., 2024). The argument here is that incremental challenges—different in scale, form, and purpose from Grand Challenges—serve as a critical organizing principle for the development of new ML and machine vision techniques (Ribes et al., 2019). Methodologically, challenges represent a fascinating setting for studying the everyday work of computer scientists working on ML model design, testing,
and application.