As we strive to develop machines that can navigate the complexities of human morality, we find ourselves confronting unsettling truths about the nature of ethics itself. Rapid advancements in AI have brought us to an era where these systems are no longer mere digital tools but key decision-makers in scenarios laden with moral consequences — from self-driving cars forced to choose between two harmful outcomes to loan-approval algorithms that determine people’s fates. This intertwining of AI and ethical challenges prompts a critical question: can we truly embed morality in AI algorithms? And what does our effort reveal about our entire ethical system? Ethics is not a single, uniform concept. It is a diverse tapestry of cultural, social, and individual threads. Standards vary significantly across societies and among individuals. Reducing this complexity to lines of code reveals profound philosophical hurdles. In “How Can AI Be Truly Ethical?” (published in The New Yorker), Paul Bloom highlights that humans themselves disagree on what constitutes moral behaviour. From a scientific perspective, classic programming is built on clear rules and logic, yet real moral dilemmas often transcend binary reasoning. The famous “trolley problem” illustrates this challenge: an autonomous vehicle must make a split-second choice without human intuition. Should it prioritise passengers over pedestrians? How does it measure one life against another? Our limited human capacity to capture the full breadth of ethics hinders our efforts to devise a clear, fair “moral algorithm” for AI. In trying to build ethical AI, we confront our own contradictions: AI systems learn from human-generated data, inevitably mirroring our biases and moral standpoints. They become a “moral mirror” reflecting both our outward behaviour and the hidden flaws in our ethical thinking. This, in turn, sparks difficult questions about responsibility. If an AI, shaped by biased data, makes a questionable decision, who should bear the blame — the developers, the users, or society at large? This dispersal of accountability introduces yet another ethical dilemma. Some argue that AI can render objective decisions free from emotional or irrational influences. However, ethics itself is not universally objective: one society’s norm may be another’s taboo. This discrepancy poses a major challenge in coding universal moral principles into AI. Moreover, a strictly “logical” AI that aims to maximise overall benefits could neglect individual rights or emotional harms, clashing with human values that attempt to balance collective and personal interests. For instance, an AI might boost workplace productivity through policies that seem efficient but result in exhausted employees or lost jobs, valuing productivity over individual well-being. As AI expands rapidly into more areas of life, robust ethical regulations become essential. Governments and international organisations must prioritise establishing guidelines to ensure AI systems operate within acceptable moral boundaries, addressing issues like privacy, consent, and the morally charged dimensions of AI-driven decisions. The EU’s General Data Protection Regulation (GDPR) represents a solid step toward creating accountability and ethical constraints. Yet laws alone cannot guarantee the ethical use of AI. Responsibility extends from the developers who build these systems to the policymakers who oversee their deployment. AI must be aligned with core ethical principles widely recognised by humanity. Some researchers propose developing explainable AI (XAI) so that the reasoning behind decisions is transparent, allowing human oversight when moral hazards arise. As AI grows more integrated into daily life, its influence on human ethics may become reciprocal. We could adopt ethical standards that mirror the logic of AI — for better or worse — shaped by digital rationality. This dynamic appears in calls to “humanise AI and digitise humanity,” hinting at a new frontier that reveals yet another moral puzzle. We may believe we can bypass the pursuit of absolute moral perfection in AI algorithms, but creating truly ethical AI involves daunting philosophical challenges, not merely technical ones. Ultimately, this pursuit highlights the fragility of our moral concepts, pushing us to reconsider what it means to be an ethical human before deciding how to code morality into machines. Indeed, AI systems serve as a moral mirror, reflecting our values and behaviours, prompting us to refine our own ethical foundations even before we impose them on intelligent algorithms.
Oman Observer is now on the WhatsApp channel. Click here
