Abstract
Since the launch of OpenAI's ChatGPT in November 2022, concerns about the spread of disinformation through "hallucinations" and the perpetuation of judgment biases by artificial intelligence (AI) have led to several international efforts to regulate AI.
Regulatory proposals for AI and other emerging media mark a turning point. According to Fred Turner, many proponents of the Web saw the potential for it to become an ideal society: "decentralized, egalitarian, harmonious, and free." Pronouncements such as John Perry Barlow’s "A Declaration of the Independence of Cyberspace" framed the Internet as a democratising space and criticises laws governments for trying to regulate it.
Although Internet has provided a plethora of benefits, largely unregulated online media such as social media platforms also became a source of privacy violations, disinformation political polarisation, radicalisation and extremism. As a result, and in contrast to their initial reluctance to regulate the internet, some governments are avoiding a hands-off approach to AI.
Governments, international organizations, academics and civil society have been addressing the question of how AI should be regulated since the mid-2010s (Kuhn, 2023). The emergence of generative AI and chatbots based on it, however, reignited interest in the possibilities and risks of these systems. Proposals to regulate AI represent two different and competing visions of the future: one characterized by human creativity, productivity, and harmony, and another characterized by polarization, conflict, and the disintegration of social cohesion.
This presentation argues that, while the media of the future might be determined by technological innovation, the future of the media will be dependent upon what media regulations governments enact, the ways they restrict media use, how well they mitigate harms and build trust in media institutions. Through a policy analysis guided by a Bardachian policy framework (see Bardach and Patashnik, 2023), and informed by a semi-qualitative "bottom-up" risk analysis (Rowe, 1991), commentary from academics, legal experts, and civil society, and the author’s own experiences as a government policy analyst, it compares AI regulatory approaches from the EU (the EU AI Act) and the UK (a "pro-innovation approach to AI regulation"). These two approaches range from voluntary measures such as industry self-regulation to regulations that impose large penalties on corporations that use AI systems that result in harm.
It is an open question whether horizontal binding regulations or vertical non-binding principles will ultimately prevail, but the approach that becomes dominant can have a major impact on the future of AI and other emergent media, in part because of the different imaginaries of the future associated with each approach. The final portion of this presentation applies the results of this analysis to the ongoing development of AI regulation in two other countries, Canada and New Zealand, to demonstrate the influence of these two different approaches and highlight issues these and other countries may want to consider as they develop their own AI regulation.