The Potential of Truth Technologies: Artificial Intelligence for Societal Resilience

Abstract

Societal resilience is fundamental for navigating sustainable development while being seriously challenged by misinformation and disinformation in the post-truth era. The thesis’ purpose is to explore how AI-driven truth technologies can be developed and utilised to mitigate mis- and disinformation. Three research questions are answered to illustrate the potential of truth technologies by (1) clarifying expectations that provide common ground and identifying promising opportunities for scaling, (2) anticipating serious risks that must be mitigated, and (3) outlining management principles to guide future development. Data is collected through 23 semi-structured qualitative interviews with distinguished professionals and academics from the field.

The results show AI is expected to be used to circulate effective disinformation, while truth technologies will become mature and widespread. Opportunities with truth technologies include improved fact-checking, evaluation of publishing sources, and the contextualisation of information. Risks are technical limitations like oversimplifications and biases, users’ overreliance, and manipulation or censorship. Management principles propose that truth technologies should be effectively and practically regulated; should be decentralised, transparent, and explainable; trustworthy bodies should oversee them; and their use complemented by education in critical thinking, with ethics considered. The discussion emphasises that truth technologies are an effective solution to false information, will be scaled, and face immediate obstacles and advanced risks. Effective management is the bottleneck to successful large-scale adoption. Further, truth technologies can be a disruptive technology, should be effectively used in hybrid models with human oversight, and underlying power structures need to be managed.

Popular Abstract

AI-driven truth technologies can fight false information and strengthen societal resilience.

The research investigates truth technologies, i.e., artificial intelligence (AI) tools that can fight false information. From 23 expert interviews, promising opportunities for the future, dangerous risks, and guiding management principles are identified.

Do you recall the time when the UK was still part of the European Union, the idea of a Donald Trump presidency was looked at amusingly, and people could agree on basic knowledge? Over the last few years, this changed more and more. Nowadays, in the “post-truth era”, there are overwhelming and confusing fake news and lies everywhere.

The thesis explores how AI can fight such false information and help societies. Concretely, such tools are called “truth technologies”. They are based on AI and not only check facts with knowledge from the internet but also analyse how people speak and what their faces express. Combining these observations, truth technologies try to decide whether a claim is true and accurate. All the following results are based on 23 interviews with experts from 11 countries from all over the world.

The results show that, on the one hand, AI is expected to be more and more used to spread false information successfully, but, on the other hand, truth technologies will improve and be used by many people. This is highly important because malicious actors use false information as a weapon to decrease mutual trust and break down democracies. The research finds opportunities with truth technologies, including their potential to improve fact-checking efforts, evaluate the trustworthiness of publishing sources, and provide more information around simplified statements. The results also warn of serious risks with truth technologies. Truth technologies could also be used by malicious actors for manipulation or censorship, they could have technical limitations, they could not be good enough to understand complexities around truth, they could be biased, and users could rely too much on them. The research presents management principles for truth technologies, and proposes that truth technologies should be effectively and practically regulated; should not be controlled by one person or organisation but decentralised and transparent; trustworthy institutions should oversee them; and the users should be supported with education in critical thinking, media literacy, and ethics.

In the discussion, the research emphasises that truth technologies are a powerful solution to fight the serious risks of false information and that the technologies will be scaled over the next years. Effective management and regulation are necessary for many people to use it well and make sure that truth technologies do not grow beyond control. Truth technologies are facing immediate obstacles that make their scaling difficult and advanced risks that appear later, once many people use them. The management principles help with both. Truth technologies might be a technology that takes everyone by surprise and grows exponentially, just like the adoption of the smartphone. The rapid progress of AI can be an early sign. Thus, understanding how humans can use these technologies is key. What’s more, there are underlying power structures that need to be managed so that truth technologies are not held in the hands of a few, but everyone has access to them, and they help societies to become more resilient.

• author: Paniego Béjar, Beatriz and Schweser, Lennart Erich
• supervisor: Magnus Hagelsteen
• organization: Division of Risk Management and Societal Safety
• course: VBRM15 20241
• year: 2024
• type: H2 - Master's Degree (Two Years)
• subject: Social Sciences; Technology and Engineering
• keywords: truth technologies, artificial intelligence (AI), post-truth era, misinformation, disinformation, fake news, societal resilience, AI opportunities, AI risks, AI regulation
• language: English
• id: 9159486
• date added to LUP: 2024-06-10 08:24:54
• date last changed: 2024-06-10 08:24:54