AI and the dilemma of rapid diffusion

The authors of this article, Filip Tichý (Partner at Grant Thornton Slovakia) and Jakub Chudík (Co-Founder at Assetario), take you through the world of artificial intelligence in the AI Breakfast series. This article was written without the use of AI.

The debate on artificial intelligence is often based on the implicit assumption that every new general-purpose technology is always implemented and adopted more rapidly than its predecessor. It took many decades for the world to become significantly electrified after Edison invented the use of electricity for lighting in 1879. The internet transformed the world beyond recognition in 10–15 years.  From this assumption, the conclusion is subsequently drawn that artificial intelligence, too, will be adopted more rapidly than previous technological paradigms, and that the so-called ‘AI race’ represents an exceptionally urgent economic and strategic phenomenon, for which we have only a few years before it is hopelessly too late. Such reasoning, however, calls for caution. As Jeffrey Funk points out in his book Unicorns, Hype and Bubbles, historically the pace of diffusion of general-purpose technologies has been determined by a number of specific factors and cannot be mechanically extrapolated to every new technological wave.

The fact that the internet and mobile phones achieved mass adoption considerably faster than cars, electrification or other older general-purpose technologies does not mean that this is a universal rule of technological development. Their accelerated adoption was the result of a specific combination of historical, infrastructural and economic conditions. The readiness of the communications infrastructure, broader post-war technological development and, above all, network effects played a significant role. The extent of the impact of the internet and mobile services on business and society grew in line with the number of users, creating a self-reinforcing pressure on the rate of diffusion and lowering the barriers to further adoption of these technologies.

In the case of artificial intelligence, it is therefore crucial to examine whether its economic value will show a similar dependence on the scale of use, or whether it will instead be a technology whose benefits are realised on an individual, sectoral or global basis. The driving force behind the global diffusion of AI could be tech giants such as Microsoft or Google, whose business model is based in part on the fact that all data is stored on a single cloud and they offer us an AI assistant or ‘Copilot’ for every aspect of our lives (both private and professional). If AI were to generate significant network effects (for example, through the dominance of its implementation by Microsoft or Google), we could expect a faster and more concentrated diffusion. However, if its use is more local or sector-specific, its roll-out may be considerably slower. The debate on the urgency of the ‘AI race’ should therefore not be based primarily on technological enthusiasm, but on an assessment of the structure and mechanisms through which artificial intelligence creates economic value and productivity. It is not yet entirely clear whether the rise of AI will be faster than that of the internet.