Title: The limits to growth for AI

Authors: Gordon Rugg; Jennifer D. Skillen

Addresses: Department of Computing and Mathematics, Keele University, Staffordshire, ST5 5BG, UK ' Department of Computing and Mathematics, Keele University, Staffordshire, ST5 5BG, UK

Abstract: This invited opinion piece is structured as follows. First, we discuss current AI in relation to the literature on innovation. We conclude that the core concepts of current artificial intelligence (AI) technologies stabilised decades ago, and are near their limits to growth. We argue that significant recent developments in AI have involved increased scale, rather than qualitative change. These increases in scale have opened up new possibilities, and it is not yet clear how far their effects will spread. One key issue is popular beliefs about AI, particularly regarding error, expert performance, and real world situations. We argue that AI successes have typically involved constrained problems, and that the structure of current AI systems is a major limiting factor. We conclude that 'AI-only' approaches face significant limits, but that a hybrid approach combining AI with human expertise has better prospects.

Keywords: artificial intelligence; growth; knowledge; human error; diffusion of innovation; expertise; knowledge infrastructure; human values; human computer interaction; HCI.

DOI: 10.1504/IJIOME.2025.148370

International Journal of Information and Operations Management Education, 2025 Vol.8 No.1, pp.61 - 73

Received: 27 Jan 2025
Accepted: 01 May 2025
Published online: 02 Sep 2025 *

Full-text access for editors Full-text access for subscribers Purchase this article Comment on this article