The Evolution and Core Conceptual Framework of AI Ethics

When Alan Turing posed the question “Can machines think?” in his seminal 1950 paper *Computing Machinery and Intelligence*, he could hardly have anticipated that the technical question of machine intelligence would soon be inseparable from the ethical dilemmas it would generate. Turing himself was already aware of the social implications of intelligent systems, warning that “we can see a short distance ahead, but we can see plenty there that needs to be done” — a prescient observation that rings truer than ever in an era where AI systems diagnose diseases, approve loans, and even make life-or-death decisions on our roads. For decades, the ethics of artificial intelligence remained a niche concern for a small group of philosophers and computer scientists, but the rapid deployment of AI in high-stakes social domains over the past 15 years has elevated it to one of the most urgent public policy and technological challenges of the 21st century. This chapter traces the evolution of AI ethics from its mid-20th century origins to its current form, clarifies its distinctiveness from both computer ethics and traditional applied ethics, lays out its core conceptual framework, compares the application of three mainstream ethical theories in AI contexts, and uses a real-world autonomous driving case study to illustrate the complexity of AI ethical decision-making. By the end of this chapter, you will have a solid theoretical foundation to engage with the specific thematic analyses of algorithmic bias, lethal autonomous weapons, and data privacy that form the core of this book.

The Historical Trajectory of AI Ethics

The history of AI ethics can be divided into three distinct phases, each corresponding to a leap in AI technological capability and societal adoption. Understanding this evolution is critical to appreciating why AI ethics is not a purely abstract philosophical exercise, but a response to concrete, real-world harms caused by automated systems.

Phase 1: The Enlightenment Era (1950–1980)

The origins of AI ethics are tightly bound to the birth of AI as a formal research field. While science fiction writers like Isaac Asimov had already explored robot ethics as early as 1942, with his iconic Three Laws of Robotics[^2], it was Turing’s 1950 paper that formally framed the question of machine intelligence as a subject worthy of rigorous academic inquiry. Turing’s work laid the philosophical groundwork for later ethical debates by establishing that machines could, in principle, exhibit behavior indistinguishable from human thought — a premise that immediately raised questions about moral responsibility for machine actions.

The first concrete AI ethics controversies emerged in the 1970s, with the development of early expert systems designed to replicate human professional judgment. The most famous of these was MYCIN, a Stanford University system developed in 1972 to diagnose blood infections and recommend antibiotic treatments. MYCIN was shown to outperform human doctors in diagnostic accuracy for certain conditions, but its deployment immediately raised ethical questions: if MYCIN made a wrong diagnosis and a patient died, who was responsible? The developers? The hospital that deployed it? The patient who consented to use an AI system? These questions had no clear answers at the time, but they foreshadowed the responsibility gaps that would become a central concern of AI ethics in the 21st century.

This early phase was characterized by a small community of researchers and philosophers exploring ethical questions in isolation, with little engagement from policymakers or the general public. The dominant view at the time was that AI ethics was a speculative, long-term concern, irrelevant to the limited, narrow AI systems that existed in laboratories.

Phase 2: Institutionalization (1990–2010)

The rise of the internet and the commercialization of AI in the 1990s brought AI ethics out of the laboratory and into the public sphere. Two key trends drove this shift: first, the widespread adoption of algorithmic decision-making in high-stakes domains like credit scoring, hiring, and criminal justice; second, the growth of data-driven surveillance and targeted advertising, which raised new questions about data privacy and algorithmic manipulation.

The 1995 EU Data Protection Directive was one of the first major policy responses to these emerging risks, establishing the principle of “informed consent” for personal data collection and the right of individuals to access and correct data held about them. While not explicitly framed as AI ethics legislation, the directive laid the groundwork for later AI-specific privacy rules by recognizing that automated data processing posed unique risks to individual autonomy.

Academic institutionalization also accelerated during this phase. The first dedicated conference on AI and ethics was held in 2000, organized by the Association for the Advancement of Artificial Intelligence (AAAI), and IEEE published its first standard for the design of autonomous systems in 2004. The 2008 global financial crisis, which was partly driven by opaque algorithmic trading systems, further raised public awareness of the systemic risks posed by unregulated AI deployment.

A key milestone of this phase was the formal distinction between AI ethics and traditional computer ethics. While computer ethics focused on the ethical behavior of human users of technology (e.g., hacking, software piracy, cyberbullying), AI ethics focused on the ethical behavior of the AI system itself as an autonomous agent capable of making independent decisions that affect human welfare. This distinction was crystallized in a 2009 paper by philosopher and AI ethicist Wendell Wallach, who argued that “the unique ethical challenges of AI stem not from the tool itself, but from the delegation of moral decision-making authority to non-human agents.”

Phase 3: The Mainstreaming Era (2010–Present)

The 2012 breakthrough in deep learning, marked by the AlexNet model’s victory in the ImageNet image recognition competition, triggered a wave of AI innovation that brought the technology into billions of people’s daily lives. The 2016 victory of DeepMind’s AlphaGo over world champion Lee Sedol in the game of Go, a task long considered impossible for machines, captured global public attention and demonstrated the potential of AI to outperform humans in complex, creative tasks.

The following years saw a series of high-profile AI ethics scandals that brought the field into the mainstream of public and policy discourse:

• The 2018 Cambridge Analytica scandal, in which Facebook user data was harvested without consent to build psychological profiles used to manipulate voters in the 2016 U.S. presidential election, exposing the risks of algorithmic manipulation and data exploitation at scale.
• The 2020 COVID-19 pandemic, during which AI systems used for disease surveillance, vaccine allocation, and lockdown enforcement were found to exhibit widespread bias against marginalized groups, including racial minorities, low-income populations, and the elderly.
• The 2022 launch of ChatGPT, which brought generative AI into the public consciousness and raised new ethical questions about intellectual property, misinformation, and the displacement of human workers.

According to the 2023 Stanford AI Index Report, the number of AI-related ethics policies enacted by national governments grew from just 1 in 2016 to 37 in 2022[^3]. The European Union’s AI Act, provisionally agreed in 2023, was the first comprehensive regulatory framework to classify AI systems by risk level and impose binding ethical requirements for high-risk applications, including bans on social scoring systems and strict transparency requirements for generative AI. The United States, China, and dozens of other countries have also released national AI ethics guidelines in recent years, marking the transition of AI ethics f