Synthetic agency: sense of agency in artificial intelligence

Artificial Intelligence and Agency: Tie-breaking in AI Decision-Making

Determining the agency-status of machines and AI has never been more pressing. As we progress into a future where humans and machines more closely co-exist, understanding hallmark features of agency affords us the ability to develop policy and narratives which cater to both humans and machines. This paper maintains that decision-making processes largely underpin agential action, and that in most instances, these processes yield good results in terms of making good choices. However, in some instances, when faced with two (or more) choices, an agent may find themselves with equal reasons to choose either - thus being presented with a tie. This paper argues that in the event of a tie, the ability to create a voluntarist reason is a hallmark feature of agency, and second, that AI, through current tie-breaking mechanisms does not have this ability, and thus fails at this particular feature of agency.

Between Meanings and Senses-Making Spaces: Agency and Ownership Emergence Formalization from Cultural-Historical Activity Theory Position, for an AI-Friendly Model

Sense of agency and sense of ownership are considered crucial in autonomous systems. However, drawbacks still exist regarding how to represent their causal origin and internal structure, either in formalized psychological models or in artificial systems. This paper considers that these drawbacks are based on the ontological and epistemological duality in mainstream psychology and AI. By shedding light on the cultural-historical activity theory (CHAT) and dialectical logic, and by building on and extending related work, this paper attempts to investigate how the noted duality affects investigating the self and "I". And by differentiating between the space of meanings and the sense-making space, the paper introduces CHAT's position of the causal emergence of agency and ownership by stressing the twofold transition theory being central to CHAT. Furthermore, a qualitative formalized model is introduced to represent the emergence of agency and ownership through the emergence of the contradictions-based meaning with potential employment in AI.

Synesthetic Correspondence: An Overview

Intramodal and cross-modal perceptual grouping based on the spatial proximity and temporal closeness between multiple sensory stimuli, as an operational principle has built a coherent and meaningful representation of the multisensory event/object. To implement and investigate the cross-modal perceptual grouping, researchers have employed excellent paradigms of spatial/temporal ventriloquism and cross-modal dynamic capture and have revealed the conditional constraints as well as the functional facilitations among various correspondence of sensory properties, with featured behavioral evidence, computational framework as well as brain oscillation patterns. Typically, synesthetic correspondence as a special type of cross-modal correspondence can shape the efficiency and effect-size of cross-modal interaction. For example, factors such as pitch/loudness in the auditory dimension with size/brightness in the visual dimension could modulate the strength of the cross-modal temporal capture. The empirical behavioral findings, as well as psychophysical and neurophysiological evidence to address the cross-modal perceptual grouping and synesthetic correspondence, were summarized in this review. Finally, the potential applications (such as artificial synesthesia device) and how synesthetic correspondence interface with semantics (sensory linguistics), as well as the promising research questions in this field have been discussed.

Configurations of human-centered AI at work: seven actor-structure engagements in organizations

Purpose

The discourse on the human-centricity of AI at work needs contextualization. The aim of this study is to distinguish prevalent criteria of human-centricity for AI applications in the scientific discourse and to relate them to the work contexts for which they are specifically intended. This leads to configurations of actor-structure engagements that foster human-centricity in the workplace.

Theoretical foundation

The study applies configurational theory to sociotechnical systems' analysis of work settings. The assumption is that different approaches to promote human-centricity coexist, depending on the stakeholders responsible for their application.

Method

The exploration of criteria indicating human-centricity and their synthesis into configurations is based on a cross-disciplinary literature review following a systematic search strategy and a deductive-inductive qualitative content analysis of 101 research articles.

Results

The article outlines eight criteria of human-centricity, two of which face challenges of human-centered technology development (trustworthiness and explainability), three challenges of human-centered employee development (prevention of job loss, health, and human agency and augmentation), and three challenges of human-centered organizational development (compensation of systems' weaknesses, integration of user-domain knowledge, accountability, and safety culture). The configurational theory allows contextualization of these criteria from a higher-order perspective and leads to seven configurations of actor-structure engagements in terms of engagement for (1) data and technostructure, (2) operational process optimization, (3) operators' employment, (4) employees' wellbeing, (5) proficiency, (6) accountability, and (7) interactive cross-domain design. Each has one criterion of human-centricity in the foreground. Trustworthiness does not build its own configuration but is proposed to be a necessary condition in all seven configurations.

Discussion

The article contextualizes the overall debate on human-centricity and allows us to specify stakeholder-related engagements and how these complement each other. This is of high value for practitioners bringing human-centricity to the workplace and allows them to compare which criteria are considered in transnational declarations, international norms and standards, or company guidelines.

Challenging presumed technological superiority when working with (artificial) colleagues

Technological advancements are ubiquitously supporting or even replacing humans in all areas of life, bringing the potential for human-technology symbiosis but also novel challenges. To address these challenges, we conducted three experiments in different task contexts ranging from loan assignment over X-Ray evaluation to process industry. Specifically, we investigated the impact of support agent (artificial intelligence, decision support system, or human) and failure experience (one vs. none) on trust-related aspects of human-agent interaction. This included not only the subjective evaluation of the respective agent in terms of trust, reliability, and responsibility, when working together, but also a change in perspective to the willingness to be assessed oneself by the agent. In contrast to a presumed technological superiority, we show a general advantage with regard to trust and responsibility of human support over both technical support systems (i.e., artificial intelligence and decision support system), regardless of task context from the collaborative perspective. This effect reversed to a preference for technical systems when switching the perspective to being assessed. These findings illustrate an imperfect automation schema from the perspective of the advice-taker and demonstrate the importance of perspective when working with or being assessed by machine intelligence.

Building and Understanding the Minimal Self

Within the methodologically diverse interdisciplinary research on the minimal self, we identify two movements with seemingly disparate research agendas - cognitive science and cognitive (developmental) robotics. Cognitive science, on the one hand, devises rather abstract models which can predict and explain human experimental data related to the minimal self. Incorporating the established models of cognitive science and ideas from artificial intelligence, cognitive robotics, on the other hand, aims to build embodied learning machines capable of developing a self "from scratch" similar to human infants. The epistemic promise of the latter approach is that, at some point, robotic models can serve as a testbed for directly investigating the mechanisms that lead to the emergence of the minimal self. While both approaches can be productive for creating causal mechanistic models of the minimal self, we argue that building a minimal self is different from understanding the human minimal self. Thus, one should be cautious when drawing conclusions about the human minimal self based on robotic model implementations and vice versa. We further point out that incorporating constraints arising from different levels of analysis will be crucial for creating models that can predict, generate, and causally explain behavior in the real world.

Foundations of Erobotics

Technology is giving rise to artificial erotic agents, which we call erobots (erôs + bot). Erobots, such as virtual or augmented partners, erotic chatbots, and sex robots, increasingly expose humans to the possibility of intimacy and sexuality with artificial agents. Their advent has sparked academic and public debates: some denounce their risks (e.g., promotion of harmful sociosexual norms), while others defend their potential benefits (e.g., health, education, and research applications). Yet, the scientific study of human-machine erotic interaction is limited; no comprehensive theoretical models have been proposed and the empirical literature remains scarce. The current research programs investigating erotic technologies tend to focus on the risks and benefits of erobots, rather than providing solutions to resolve the former and enhance the latter. Moreover, we feel that these programs underestimate how humans and machines unpredictably interact and co-evolve, as well as the influence of sociocultural processes on technological development and meaning attribution. To comprehensively explore human-machine erotic interaction and co-evolution, we argue that we need a new unified transdisciplinary field of research-grounded in sexuality and technology positive frameworks-focusing on human-erobot interaction and co-evolution as well as guiding the development of beneficial erotic machines. We call this field Erobotics. As a first contribution to this new discipline, this article defines Erobotics and its related concepts; proposes a model of human-erobot interaction and co-evolution; and suggests a path to design beneficial erotic machines that could mitigate risks and enhance human well-being.

A Bayesian psychophysics model of sense of agency

Sense of agency (SoA) refers to the experience or belief that one’s own actions caused an external event. Here we present a model of SoA in the framework of optimal Bayesian cue integration with mutually involved principles, namely reliability of action and outcome sensory signals, their consistency with the causation of the outcome by the action, and the prior belief in causation. We used our Bayesian model to explain the intentional binding effect, which is regarded as a reliable indicator of SoA. Our model explains temporal binding in both self-intended and unintentional actions, suggesting that intentionality is not strictly necessary given high confidence in the action causing the outcome. Our Bayesian model also explains that if the sensory cues are reliable, SoA can emerge even for unintended actions. Our formal model therefore posits a precision-dependent causal agency.

Sense of agency (SoA) refers to the experience that one's own actions caused an external event. Here, the authors present a model of SoA in terms of optimal Bayesian cue integration taking into account reliability of action and outcome sensory signals and judging if the action caused the outcome.