The Debate Has Moved. Here's Where It Stands Now.

We already covered the fundamental distinction between AI processing information and actually thinking. That argument is settled enough. What's changed is who's now weighing in — and what they're demanding.

Three very different institutions dropped new material on AI's limits and obligations almost simultaneously. UNESCO updated its AI ethics framework. Microsoft published its 2025 Responsible AI Transparency Report. And the New York Times ran a piece asking whether Kant — yes, the 18th-century German philosopher — holds the key to understanding what AI fundamentally cannot do.

They're all circling the same drain. None of them have a clean answer.

UNESCO's 2021 Standard Is Now Being Enforced. Sort Of.

Back in November 2021, UNESCO adopted what it called the first-ever global standard on AI ethics — the Recommendation on the Ethics of Artificial Intelligence. It applies to all 194 UNESCO member states. The document was last updated September 26, 2024, according to UNESCO's own site.

The Recommendation's stated cornerstone is human rights and dignity. It covers data governance, environmental impact, gender, education, health, and social wellbeing. Big list. Ambitious scope.

A UNESCO "recommendation" carries ZERO binding legal force. Member states can ignore it entirely. There is no enforcement mechanism. No fines. No sanctions. It's a strongly worded suggestion signed by governments that include China and Russia.

UNESCO is also now running a Global AI Ethics and Governance Observatory — described as a resource for policymakers, regulators, academics, and the private sector. Good idea on paper. But an observatory that watches things happen is not the same as a body that stops bad things from happening.

Microsoft Says It's Being Responsible. Here's What They Actually Published.

Microsoft released its 2025 Responsible AI Transparency Report, outlining six core values the company claims guide its AI development: fairness, reliability and safety, privacy and security, transparency, accountability, and inclusiveness.

Those are good words. Every one of them. The practical question is how you measure any of that.

Microsoft's own materials flag an "AI divide" — a gap between populations that have access to AI tools and those that don't. Their research arm published pieces on "advancing AI to meet the needs of the global majority" and a separate AI Diffusion Report tracking where AI is most used, developed, and built.

The answer is concentrated in wealthy Western nations and China. The rest of the world is mostly on the receiving end of decisions made elsewhere.

Microsoft deserves credit for publishing transparency reports at all — most tech companies don't. But a company self-grading its own ethics practices is not the same as independent accountability. The 2025 Digital Defense Report they also referenced covers cybersecurity threats from AI. That's real and urgent. It got less attention than the feel-good inclusiveness language.

The NYT Asks If Kant Can Save Us

The New York Times piece, titled "What A.I. Kant Do," offers the most intellectually honest assessment of the three.

The argument, loosely: Immanuel Kant's moral philosophy rests on rational autonomy — the idea that moral agency requires genuine reasoning, not pattern matching. AI doesn't reason. It predicts. Those are not the same thing.

The NYT frames this partly as good news for humanities graduates, which is cute. But the underlying point is serious. If AI systems cannot exercise genuine moral judgment — and the evidence strongly suggests they cannot — then handing them decision-making authority in courts, hospitals, or military systems isn't just risky. It's philosophically incoherent.

Coverage treated the piece as a quirky culture story about English majors. It's actually a direct challenge to the entire "responsible AI" framework being sold by UNESCO and Microsoft alike. You cannot make an AI "accountable" in any meaningful sense. Accountability requires agency. Agency requires genuine reasoning. We're not there.

What Everyone Is Getting Wrong

Left-leaning outlets like the NYT focus on bias, inequality, and who gets harmed — legitimate concerns, but they consistently underweight the national security dimension. China is building AI with ZERO interest in UNESCO's ethics recommendations. That asymmetry doesn't get nearly enough ink.

Tech companies like Microsoft frame responsibility in terms of internal principles and voluntary transparency. That's better than nothing, but it's also exactly what you'd say if you wanted to avoid government regulation.

UNESCO represents the multilateral consensus view — comprehensive, well-intentioned, and largely toothless without enforcement teeth.

The philosophical limits of AI reasoning, the geopolitical race to deploy it anyway, and the governance vacuum where real accountability should be — these are not being connected.

What This Means for Regular People

If you're a worker, a patient, a student, or a citizen — AI is being embedded into systems that affect your life right now. The institutions claiming to govern it disagree on basic definitions, have no shared enforcement power, and are racing to appear responsible without being held to a binding standard.

The philosophy is real. The ethics frameworks are real. The accountability is not.

Until someone builds a governance structure with actual consequences for violations — not recommendations, not transparency reports, not op-eds about Kant — the people making the rules are grading their own homework.