Importance and limitations of AI ethics in contemporary society

The paper examines the importance and inherent limitations of AI ethics amid rapid advances in robotics, artificial intelligence, and machine learning. It frames the research question as how contemporary ethical frameworks and legal regimes can and should respond to increasingly autonomous AI systems that permeate social, economic, and political life. Using emblematic cases (e.g., the humanoid robot Sophia’s citizenship and public presence) and policy developments (particularly in the EU), the article highlights why ethical scrutiny is urgent, what issues recur (e.g., autonomy, explainability, accountability, personhood), and how societal values should inform the design and deployment of AI. The purpose is to map current approaches, identify unresolved ethical-legal dilemmas, and underscore the need for trustworthy, human-centric AI. The article stresses the significance of preparing for AI’s societal impacts through robust ethical guidance that complements technical robustness.

The article synthesizes strands from machine ethics, robotics ethics, AI policy, computational creativity, and technology law. It references foundational discussions on machine and robot ethics (Anderson & Anderson; Allen et al.; Moor; Wallach; Lin et al.; Boddington; Boden; Brynjolfsson & McAfee; Bostrom), and the emergence of human-centric, trustworthy AI in EU strategy documents (European Commission Communications, High-Level Expert Group, European AI Alliance). It situates ethical concerns within broader cultural and conceptual narratives (Domingos; Dormehl; Oliveira) that frame algorithms as central to contemporary understanding. It surveys literature on story generation and computational creativity (Gervas; Callaway & Lester; Labbé & Labbé), automatic text generation and its limits (e.g., Parker’s automated books; SCIgen), and debates on copyright and authorship of computer-generated works across jurisdictions (Bridy; Barbosa), noting divergent legal treatments in the US, UK, and New Zealand. It also draws on legal and policy analyses of autonomous vehicles and liability (Marchant & Lindor; Scherer; Garza; Vladeck), empirical preference elicitation via MIT’s Moral Machine, and national ethics guidance (German Ethics Commission).

Conceptual and normative analysis supported by illustrative case studies and policy review. The author synthesizes existing academic literature, legal and policy documents (notably from the EU), publicized events (e.g., Sophia’s citizenship, the 2018 Uber self-driving fatality), and thought experiments (e.g., Flatland analogy; traveling salesman problem) to surface ethical tensions. The approach is descriptive-analytical rather than empirical: it identifies issues, contrasts frameworks (e.g., utilitarian vs. deontic priorities in crash dilemmas), and discusses regulatory options (e.g., strict liability, electronic personhood) to highlight both possibilities and constraints of AI ethics in practice.

• AI ethics is indispensable but inherently limited when confronting rapidly evolving, autonomous AI systems that introduce novel legal and moral challenges (e.g., personhood, accountability, value alignment).
• EU frameworks foreground Human-Centric and Trustworthy AI, emphasizing respect for fundamental rights and technical robustness. Trustworthy AI requires: (1) ethical purpose aligned with EU values and rights; (2) technical reliability to avoid unintended harm. Ethical requirements should be integrated across the AI lifecycle.
• Publicity events like Sophia’s citizenship crystallize debates on robot rights and potential electronic personhood, highlighting risks of disrupting traditional legal notions of personhood and accountability.
• Computational constraints (e.g., combinatorial explosion illustrated by the traveling salesman problem: 25 cities entail over 1.55×10^25 possible routes) show intrinsic limits of computational problem-solving relevant to expectations of AI capability.
• Automated content generation is widespread (e.g., Philip M. Parker’s system has produced over 200,000 books; newsroom bots like Quakebot and Quill), raising questions about creativity, originality, and authorship. Jurisdictions diverge: in the US, the user is typically author; in the UK/NZ, the person who undertakes the necessary arrangements is recognized when no human author exists.
• Autonomous vehicle ethics exemplify the distribution of unavoidable harm: society’s preferences are inconsistent. Moral Machine results suggest preference for saving younger over older individuals, but no consensus on prioritizing occupants vs. bystanders; consumers favor utilitarian programming in principle but resist buying such cars for themselves.
• Accident and liability considerations: humans cause about 90% of serious road accidents; ~1.3 million deaths occur annually worldwide from road traffic incidents, suggesting potential benefits from autonomy, yet responsibility allocation remains unresolved across mixed-control and fully autonomous scenarios.
• Regulatory directions include using existing vehicle liability rules for driver-assisted systems, product liability for algorithmic defects, and proposing strict civil liability for manufacturers for fully autonomous decisions. Proposals for electronic personhood are controversial and premature.
• Germany’s Ethics Commission recommends: prioritizing human life over animals/property; prohibiting discrimination by age, gender, or health in dilemma settings; and permitting configurations that minimize the number of injured. Government-led frameworks may guide programming despite public ambivalence.
• Attempts to encode moral decision-making (e.g., inverse reinforcement learning; artificial moral agents) face conceptual and methodological contradictions; ethics guidance must acknowledge persistent dilemmas and societal value pluralism.

The analysis shows that while AI ethics frameworks provide essential guardrails—centering human rights, accountability, and safety—they also encounter practical and theoretical limits when translated into code and policy. Cases like Sophia expose conceptual strains in extending legal constructs (personhood) to artificial entities, risking diffusion of responsibility. Autonomous vehicles reveal a gap between public ethical ideals and consumer behavior, complicating market and regulatory solutions. EU strategies demonstrate how governance can shape trustworthy AI, but cross-border differences and evolving capabilities demand adaptable, context-sensitive policies. Computational and creative examples illustrate that both overestimation of AI’s capacities (ignoring hard limits) and underestimation (discounting socio-technical impacts) can misguide ethics. Ultimately, the findings argue for iterative, multi-stakeholder ethics integrated across the AI lifecycle, coupled with clear liability regimes (favoring manufacturer responsibility over speculative electronic personhood), transparency about system behavior, and caution in delegating moral judgments to algorithms. This combined approach addresses the research aim by clarifying where ethics is most impactful (design, deployment, governance) and where its limits necessitate legal and institutional complements.

Understanding how AI functions and how it reshapes moral responsibility will be essential for all who interact with it. AI increasingly undertakes human activities, yet development remains in human hands and should be steered toward human benefit, balanced by caution. To prepare for phenomena such as superintelligence, artificial consciousness, and accelerated AI growth, the field of AI ethics must be further developed to model potential impacts on responsibility and trustworthy AI. The paper concludes that ethics must be embedded throughout AI’s lifecycle, complemented by robust legal frameworks (e.g., clear liability), and informed by ongoing societal dialogue—recognizing that some dilemmas will persist and require continuous refinement rather than one-time solutions.

The paper is a conceptual synthesis without original empirical data; thus, conclusions rely on secondary sources, illustrative cases, and policy documents. Ethical preference data (e.g., Moral Machine) may not generalize across cultures or to real-world behavior. Regulatory analysis is Eurocentric, with emphasis on EU and German contexts, limiting applicability elsewhere. Some technologies discussed (e.g., fully autonomous vehicles at scale, electronic personhood) are prospective, so recommendations address evolving targets. The acknowledged inconsistency of societal ethical positions constrains the prospect of universally acceptable algorithmic solutions.