Transdisciplinary AI Observatory -- Retrospective Analyses and Future-Oriented Contradistinctions

Abstract

In the last years, AI safety gained international recognition in the light of heterogeneous safety-critical and ethical issues that risk overshadowing the broad beneficial impacts of AI. In this context, the implementation of AI observatory endeavors represents one key research direction. This paper motivates the need for an inherently transdisciplinary AI observatory approach integrating diverse retrospective and counterfactual views. We delineate aims and limitations while providing hands-on-advice utilizing concrete practical examples. Distinguishing between unintentionally and intentionally triggered AI risks with diverse socio-psycho-technological impacts, we exemplify a retrospective descriptive analysis followed by a retrospective counterfactual risk analysis. Building on these AI observatory tools, we present near-term transdisciplinary guidelines for AI safety. As further contribution, we discuss differentiated and tailored long-term directions through the lens of two disparate modern AI safety paradigms. For simplicity, we refer to these two different paradigms with the terms artificial stupidity (AS) and eternal creativity (EC) respectively. While both AS and EC acknowledge the need for a hybrid cognitive-affective approach to AI safety and overlap with regard to many short-term considerations, they differ fundamentally in the nature of multiple envisaged long-term solution patterns. By compiling relevant underlying contradistinctions, we aim to provide future-oriented incentives for constructive dialectics in practical and theoretical AI safety research.

Facing Immersive “Post-Truth” in AIVR?

“…VR could thus represent a suitable awareness-raising tool for future severe AI(VR) safety risks, such as by facilitating valuable retrospective counterfactual analyses [72]. Fourth, a generic recommendation that may already be applicable nowadays is to deliberately turn the confirmation bias [73] automatically reinforced via AI-empowered social media [74] against itself [57]. For example, one could create social media spaces (subsuming future social VR) that reinforce critical thinking, life-long learning, and criticism [57], which could be deliberately fueled via artificial bots (or non-player characters in VR), steering attention towards those patterns.…”

“…Fourth, a generic recommendation that may already be applicable nowadays is to deliberately turn the confirmation bias [73] automatically reinforced via AI-empowered social media [74] against itself [57]. For example, one could create social media spaces (subsuming future social VR) that reinforce critical thinking, life-long learning, and criticism [57], which could be deliberately fueled via artificial bots (or non-player characters in VR), steering attention towards those patterns. Even if immersive falsehood would often not be resolved quickly, (AI-aided) social peer pressure reinforcing critical thinking and a focus on invariant good explanations could represent a necessarily incomplete, but principled defense.…”

“…For instance, future immersive falsehood in the form of AI-manipulated VR news for disinformation [22], but also defamation and extortion purposes, could distort historical and forensic records and exacerbate issues in the information ecosystem. Malicious actors could craft future realistic immersive experiences (e.g., of fake AI-generated confirmatory experiments and research [57]) to undermine the scientific enterprise. With increasing degrees of realism, many scientists may not stay immune against such strategies.…”

“…By contrast, Popperian epistemology [55] sees science as an explanation-based and criticism-centered endeavor with falsifiability as decisive criterium-which has been extended by Deutsch [56] who views science as the quest to identify invariant hard-to-vary explanations of reality. On that view, deepfakes (and immersive falsehood) do not put truth at risk (see [57] for more details including the safety-relevant urgency to thematize these fundamental Bayesian vs. Popperian epistemic divergences).…”