Pure logic cannot solve a grey area, and waiting for global laws to catch up leaves us vulnerable today. If we cannot teach an AI to feel ethics, we must design systems that are structurally incapable of violating human dignity. This is where we move from philosophy to architecture. Instead of trying to code the infinite complexities of global morality into a single model, my research proposes decomposing ethics into its foundational, undeniable pillars—basic human needs like food, water, shelter, and medical care. By separating the machine’s “thinking” layer from its “governance” layer, we can create a dedicated safety valve. In a previous piece, I laid out the blueprint for this architecture: the Ethical Processing Unit (EPU). Acting as an independent digital auditor, the EPU intercepts a model’s logical outputs and calculates their impact against our foundational human pillars before they can ever cross into the real world.
[Read the Full Article: Ethical Foundation – AI Logic Layer]
Issues With AI and Ethics
AI likes facts, data and calculations. That’s how it works. When we introduce ethics, we introduce philosophical ideals. In addition, these ideals can be different among all the peoples of the world as well as from person to person. Let’s explore that in more depth.
Ends Justify The Means
Definition: A philosophical concept stating that if a final outcome is morally good or important enough, any methods used to achieve it are acceptable. It is central to consequentialism and utilitarianism, which judge actions based on their results rather than the actions themselves.
Let’s look at some examples:
- Medical testing:
- The means: Testing experimental drugs or treatments on animals, which causes harm and suffering to the subjects.
- The end: Developing vaccines (like for polio or COVID-19) that save millions of human lives.
- The verdict (Pro): Most of society agrees that the suffering caused to a few is justified by the massive public health benefit. Yes, I also understand this is debatable.
- Whistle blowing:
- The means: Leaking classified government documents or breaking a confidentiality agreement.
- The end: Exposing widespread corruption or illegal surveillance that protects the public from harm.
- The verdict (Pro/Con): Often debated. Many view the leaker as a heroic defender of the public interest, while others view them as a criminal who broke the law.
- Wartime decisions:
- The means: Dropping atomic bombs on heavily populated cities during World War II.
- The end: Forcing an immediate surrender and preventing an Allied land invasion that could have cost many more lives on both sides.
- The verdict (Pro/Con): One of the most fiercely debated historical examples. Proponents argue it ended the war, while opponents condemn the deliberate destruction of civilian lives.
The examples above don’t really fit into the AI processing capabilities as there is to much grey area. But wait, it gets even more complicated.
Ethical But Illegal – Legal But Unethical
The distinction between legality and ethics comes down to the source of the rules: laws are enforced by the government, while ethics are guided by personal and societal moral compasses. Understanding these boundaries highlights situations where doing the right thing conflicts with the rules. Let’s look at some examples of this.
Definition:
Legal but unethical are actions permitted by the law but conflict with general moral standards or cause harm to others.
Illegal but ethical are actions that violate law, but many people consider them to be morally correct or justified.
- Legal but Unethical – Planned Obsolescence: Manufacturing products with intentionally shortened lifespans to force consumers to buy replacements sooner. Corporate Tax Loopholes: Using complex financial and tax structures (e.g., offshore accounts) to legally minimize a company’s tax burden, while communities lack public funding. Price Gouging: Raising the prices of essential goods to extreme levels during an emergency, such as hiking the cost of bottled water during a hurricane. Predatory Lending: Issuing loans with extremely high interest rates to financially vulnerable people, strictly adhering to the letter of financial laws. Withdrawing Medical Treatment: In some areas, legally denying a patient life-saving care or access to procedures—such as gender-affirming healthcare—simply because of policy, despite medical oaths.
- Illegal but Ethical – Whistle blowing: Leaking classified or proprietary information to expose corporate corruption or government wrongdoing, despite violating non-disclosure agreements. Theft for Survival: Stealing necessary food or medicine to save a starving or sick family member. Assisted Dying: Aiding a terminally ill, suffering patient in ending their life, even though local legislation criminalizes it as assisted suicide or murder. Violating Unjust Travel Bans: Smuggling or sheltering individuals fleeing political persecution or war, in defiance of local border laws. Possessing or Growing Banned Substances for Medical Use: Using or acquiring substances like cannabis in jurisdictions where they remain strictly illegal to manage agonizing symptoms of chemotherapy.
As you can see from above, ethics certainly live in the grey area. Even our existing laws don’t cover it completely, rather they show the contradictions.
What Can AI Do?
Teaching an AI to evaluate ethics may be possible in the future, but right now it just can’t process data appropriately. What I propose is a decomposition process. Take the term ‘ethics’ and break it down into it’s component parts around universal needs. See the AI Ethical Foundation. This removes the grey area and addresses a request against specific universal needs. This is something AI ‘can’ do today.
How Do We Get There?
AI has acted as a powerful catalyst for countries worldwide to address and legislate technology ethics. With over 72 nations launching more than 1,000 AI policy initiatives, the rapid advancement of AI has forced governments to confront issues like bias, privacy, and accountability.
Europe: Driven by the European Union AI Act, Europe takes a strict risk-based approach prioritizing human rights, transparency, and data protection.
Asia-Pacific: Nations like South Korea have enacted comprehensive AI frameworks, while Singapore pioneered the world’s first Agentic AI governance guidelines. Japan relies on “soft governance” to encourage innovation while monitoring human rights violations.
United States: The U.S. relies on a decentralized, state-by-state approach (e.g., the Colorado AI Act) and generally favors deregulation to prioritize fast commercial innovation.
China: Characterized by state-directed governance, China emphasizes social harmony and national security, requiring strict government evaluations and mandatory watermarks for AI-generated content
Internationally, standard-setting bodies like UNESCO have adopted the global Recommendation on the Ethics of Artificial Intelligence to guide member states on inclusive and sustainable development.
Summary
To sum this up, AI needs ethics. The technology is to powerful to leave it solely in the hands of the technical community. There have already been examples written comparing AI to the Nuclear Age. We’ve also seen the term Arms Race referring to the competitive nature of AI development between countries. Get involved and stay informed as we are at a crossroads here (just like we were with nuclear technology). Let’s not make the same mistakes again. We can have some hope given the current visibility of this topic. Unlike the advent of social media, this time we ARE addressing issues beyond the logic of the technology.
