Disruptions from Outpaced Societal Adaptation

"Though there is uncertainty among experts about the exact scale of impact that increasingly advanced general purpose AI models could have, some experts believe that these models can be compared to other general purpose innovations like the steam engine, the railroad or electricity.215 While the advent of these innovations had a significant positive effect during the industrial revolution, the widespread adoption of new technology usually comes with some level of disruptive consequences to societies. The speed and scale at which general purpose AI models are currently being adopted might not allow for much time to understand and react to societal disruptions."(p. 40)

"Even those with optimistic predictions about the impacts of AI on the labour market warn that society may lag in adapting to the rise of AI at the workplace, thus missing out on implementing re-skilling or social safety mechanisms, and thus potentially increasing wage inequality.216"(p. 40)

"While there is uncertainty about the magnitude of labour market effects caused by AI, there are certain novelties about the potential disruptions of general purpose AI.217 For the first time, developments in AI technology could replace ”high-skill“ or ”knowledge” jobs218, including in creative fields such as music, art, and journalism, or customer service or administrative roles219. Ambiguity surrounding the copyright protection of training data and AI-generated creative outputs poses additional challenges in fair compensations for original creators, especially because general purpose AI models can easily recreate another artist’s style.220"(p. 41)

"The risks from societal disruptions caused by general purpose AI are not limited to the workforce, but also extend to areas like the education system.221 If adoption of ever more capable AI models keeps outpacing educational institutions, numerous challenges could arise. Initially, general purpose AI powered tools like OpenAI’s ChatGPT were quickly banned in educational institutions due to fears of plagiarism and hampering critical thinking, struggling to distinguish student- and AI-generated work.222 In contrast, the European University Association already advocated for a more adaptive than reactive approach in order to effectively use this technology.223 A lack of proper instruction for the effective use of and knowledge about AI technology might leave students ill-prepared for a rapidly changing job market.224 Furthermore, as general purpose AI models become increasingly integrated into the educational process as personalised tutors225— already piloted in applications like Duolingo226 or Khan Academy227 — issues around accessibility, equity and loss of genuine human interaction228 in teaching need to be addressed."(p. 41)