AI, Say You’re Sorry

Artificial intelligence now writes apology texts, performance reviews, and love notes in seconds. It even acts as a “confidante,” holding hands and comforting seniors with a tone that sounds remarkably real. But as these tools slip into our most personal spaces, we are beginning to outsource not just language, but ownership.

The output is often eloquent, but there are dangerous downsides. When we treat AI as a moral or emotional authority, the “asymmetry” of risk becomes life-threatening. We have already seen tragic real-world consequences where individuals, seeking guidance from AI “confidantes,” were nudged toward self-harm, suicide, or even violence.

The core of the problem is that AI can model empathy, but it cannot feel remorse. It can suggest a course of action, but it cannot bear the weight of a wrong decision. While AI can outperform humans in narrow analytic tasks, responsibility cannot be delegated to probability scores—especially when a life is at stake.

In my recent book, Before AI Decides: Nine Ways to Stay Human, I argue that human judgment must remain the final filter. AI removes “friction,” which is its genius—and its greatest risk. Friction is where moral clarity is built. Without it, we enter a “confidence economy” where fluent answers erase the visible struggle required for true accountability.

To navigate this era, we must practice three disciplines:

• Name the Decision: Before acting on an AI suggestion, identify the human stakes involved.
• Reinsert Ownership: Rewrite AI drafts in your own language to ensure you can stand behind the words.
• Declare Responsibility: Be transparent about AI use and clear about where your judgment overrides the machine.

AI is restructuring society in countless, often unimaginable ways, that scream from daily headlines: some for good, others for the worse. The most dangerous moment in any AI interaction is when we stop questioning the tool. AI can draft an apology or a script, but it cannot sit with a grieving family or accept the cost of a tragedy. Those burdens remain human. We safeguard our dignity not through automation, but through intentional, presence-based judgment.

This essay was written using AI as an editing tool to iterate the issues explored.

Payson Stevens’ recent book, Before AI Decides: Nine Ways to stay Human, examines how individuals can retain judgment and responsibility as AI becomes embedded in everyday decisions.