AI and The Fifth Estate: The Power Shift of the Digital Age

The emergence of ChatGPT and related Artificial Intelligence (AI) tied to Large Language Models (LLMs) is the hot-button issue of the day. There are those who see AI as the new-new thing that will boost productivity and jobs. Others have raised concerns over LLMs being error prone – like other media, in not getting the facts right, and still others – including some highly respected AI experts and executives – stoking fears over some future machine learning powered AI becoming so powerful that it will put jobs and even humanity at risk. Still others doubt the hype around AI, believing it will not have a remarkable impact with the expert forecasts being predictably wrong.

Despite decades of research and development on AI, the sudden rise of ChatGPT since November of 2022 kicked off an unprecedented fear campaign to promote regulatory restrictions on innovation of AI, with its impressive functionality fueling futuristic visions of generative AI someday surpassing human intelligence. My own view is that the Utopian and dystopian forecasts are exaggerated and that the advances it can deliver in the near-term, despite well-known weaknesses, are sufficient to advance existing functionalities, such as search. Moreover, these capabilities can and should move into the hands of ordinary people who can use applications like ChatGPT strategically and responsibly to enhance the Fifth Estate power shift of the digital age.

At this point, the dystopians of AI are dominating the news in the US and Europe, with calls for new policy and regulation to pause its development and application. However, ideas for putting up guard rails or limiting access of LLMs to a large corpus of material, such as through stronger copyright restrictions, could derail promising innovations for ordinary internet users while also putting us behind bad actors that do exploit advances in AI. However, the dystopian critics have been standing on the shoulders of pioneers in this field, such as the godfather of AI, Geoffrey Hinton. Their views should be taken seriously but with the understanding that technologists are often too focused on following the logic of its technical features. They know less about how most people use the internet, social media, and related technologies, like search, and how they will use AI. Arguably, the potential of applying AI in applications like ChatGPT are apparent and need to be prioritized and widely diffused. They can support the empowerment of ordinary internet users like you and me. How so?

In my book, The Fifth Estate, I argue that the internet has already enabled the empowerment of networked individuals through their strategic use for searching, creating, sharing, collaborating, and leaking information online. As in the case of the press enabling a Fourth Estate in an earlier era, by enabling networked individuals to strategically employ these functionalities, the internet has enabled a Fifth Estate that is institutionally and constitutionally different from other estates of the political realm, such as the press, and is relevant to enhancing social accountability across all sectors of society.

While estate theory was a notion of the different sources of power in a pre-industrial society, it in fact continues to capture fundamental aspects of pluralistic power structures in democratic societies of today. The Fifth Estate can support greater pluralistic accountability in politics and society by enabling networked individuals to check the other estates, including the press and other institutions. This is not only a descriptive theory, based on research reviewed in my book, but also a normative prescription of how to support more pluralistic democracies through greater political and social accountability.

One of my principle aims in writing this book was to help create a better understanding of the existence and potential of this Fifth Estate. If it is better understood, then networked individuals could become more self-aware of their agency and capable of more effective action. In the case of AI and LLMs, networked individuals should be able to see that these applications, like OpenAI’s ChatGPT, particularly as they arrive on their laptops and mobile devices, will empower them to create content, search, and collaborate more seamlessly and effectively – empowering their role as a Fifth Estate.

One of the true pioneers of the Internet, Douglas Engelbart, argued that AI is better defined as augmented rather than artificial intelligence. Computer and related information and communication technologies like the internet and machine learning are advancing tools that augment the capabilities of networked individuals as well as institutions, like the press, rather than simply becoming a substitute for human intelligence. This is exactly the case with applications like ChatGPT. They enable networked individuals to conduct such functions as search more rapidly and across a wider corpus of text. By reading the web, as a corpus of text, there are inevitable risks of garbage in, garbage out (GIGO) – but there is no corpus of the truth. Despite problems and limitations with these applications, these are not unlike those with other sources of information and represent challenges for the developers and users of AI.

The results provided by such applications require individuals to check and authenticate the results, as has always been the case. The most valuable initiatives for policy and practice should be focused on educating the public not only in multimedia literacy but – how to read and view content critically, how to check sources and claims, and to be sceptical of claims whether they are in line or opposed to their personal beliefs and preferences. The issue of people confirming their pre-existing biases online is not a technological problem as the internet brings more diverse sources of information to your fingertips. Confirmation bias is a psychological propensity that people can be made aware of and taught to counter by using the internet and social media to access and consider a diverse array of facts and opinions that can challenge rather than simply confirm prevailing views on any issue more easily. My recommendation is that you get experience in using applications like ChatGPT before you decide on its benefits and risks. Social scientists and other internet researchers should be focusing on the study of how such applications are being used – the technical experts need to know.

Bibliography

William H. Dutton (2023), The Fifth Estate: The Power Shift of the Digital Age. New York: Oxford University Press.

Yorick Wilks (2019), Artificial Intelligence: Modern Magic or Dangerous Future? London: Icon Books.

Comments are most welcome