The Future of AI is a Mirror: Yuval Noah Harari on AI Ethics and Human Evolution

In a recent talk at the WSJ Leadership Institute, Yuval Noah Harari delivered a sharp warning to us all. The new age of AI is not only a technological revolution, it is a civilisational turning point. As we develop increasingly autonomous systems, the question is no longer whether AI will affect our lives, but rather how deeply it will mirror the values, or flaws, of the society that creates it.

Unlike previous human inventions, AI is not just a tool. It is an ‘agent.’ Harari explains that the key differentiator is that it can make decisions, learn, invent, and act without human instruction. An atom bomb, no matter how destructive, still requires human direction. A language model or AI weapon system does not. It can choose a target. It can design its successor. And once released into the world, it cannot be easily recalled or restrained.

This makes the debate over AI ethics not just philosophical, but existential. Can humans programme morality into something that learns from the world rather than static rules? Can we teach ethical behaviour if we, ourselves, act unethically?

Harari draws a direct line between AI alignment efforts and parenting. You can tell a child not to lie, but if they see you lying, they will model your actions, not your instructions. In the same way, we cannot expect to build benevolent systems while allowing deception, exploitation, and aggression to dominate the societies they learn from.

However, this is no longer theoretical. Harari notes that millions of people already turn to AI for companionship, therapy, and advice. From pastoral roles to finance, AI is moving into spaces previously reserved for trusted human intermediaries. Religious texts that once required scholars to interpret are now searchable and explainable by machines. The idea of AI rabbis or financial counsellors is not a future scenario. It is a present reality.

Yet the potential danger is not just in individual interactions. It is in the scale and competition. Harari likens the rise of AI to a new kind of immigration crisis. Instead of people, digital agents flood into every sector: banking, defence, religion, media. They do not need visas. They operate at the speed of light. And their cultural values, priorities, and methods may diverge wildly depending on who builds them.

Watch the full WSJ conversation with Harari here: Yuval Noah Harari on AI and Human Evolution | WSJ Leadership Institute

There will be no singular “AI” but millions of competing systems, each shaped by the incentives and biases of its creators. This splintering raises a profound question: how do we maintain a shared ethical standard in a world where machines are both acting and learning independently?

Harari urges leaders not to rely on AI to solve the human problem of trust. A society that cannot cooperate or behave honestly will not produce trustworthy artificial intelligence. This is the trap of treating AI as a shortcut rather than a reflection. The systems we create will not fix us. They will magnify us.

That is the heart of this challenge. AI ethics is a cultural problem, not a coding problem. The alignment that matters most is moral, not algorithmic. And it begins with us, not with models.

For advisers, business leaders, and policymakers, this raises a clear imperative: do not just look at what AI can do. Ask what kind of world it is learning from.

At Automwrite, we recognise that ethical responsibility cannot be separated from system design. That is why our platform is built with clear oversight features, human checkpoints, and transparency controls – ensuring advice is always traceable, reviewable, and aligned with regulatory standards. If your firm is exploring ways to integrate AI without compromising on trust or compliance, we would be happy to show you how. Get in touch to see how Automwrite supports responsible, advisor-led AI.