Fairness and AI: Why We Build Merciful the Way We Do

If you had to design artificial intelligence without knowing who you’d be in the world, what kind of software would you build?

October 2, 2026 · 5 min read · by the Merciful team

If you have ever seen two siblings split the last slice of birthday cake, you already know the oldest trick in parenting: “You cut, I choose.”

If you let one child cut the cake and also pick the slice, the result is predictable: they take 90% of the frosting for themselves and slide a plate of crumbs to their brother. But when the other person gets first pick, the child with the knife suddenly measures every millimeter with surgical precision. They know that if they cut an unfair piece, they will be the one stuck eating it.

In 1971, Harvard philosopher John Rawls published A Theory of Justice. He took that exact cake-cutting intuition and turned it into one of the most famous thought experiments in modern ethics: the Veil of Ignorance.

Rawls asked: How do we know if an institution or system is genuinely fair?

His answer was to imagine that everyone sits down to design the rules of the world, but right before anyone speaks, a magical blindfold drops over everyone’s eyes. Behind this blindfold, you don’t know who you will be when you step out into reality:

• You don’t know if you will be rich or living paycheck to paycheck.
• You don’t know if you will be a Silicon Valley executive with clusters of GPUs, or an entry-level worker whose job just got replaced by a script.
• You don’t know if you will be a seasoned machine-learning engineer or a student in a rural classroom with spotty Wi-Fi.
• You don’t know your health, your background, or your social standing.

Behind that blindfold, you don’t know which slice will end up on your plate. You cannot afford to cut carelessly. Instead, you design the system so that even the person who gets the very last pick has a dignified, secure, and empowering experience.

Who Holds the Knife Today?

Now look at the artificial intelligence industry in 2026.

The people deciding how AI works today are not wearing a blindfold. They are holding the knife, and they already know they get the first pick. They control the GPU data centers, the venture capital, the proprietary datasets, and direct lines to regulators.

Because they already know they get the biggest slice, the default playbook in consumer AI often feels backwards for regular people:

• Optimizing for addiction over utility: Apps are deliberately engineered to simulate affection and human attachment, hooking lonely or vulnerable people to drive daily active usage and subscription renewals.
• Toll booths on basic thought: Advanced reasoning assistance is increasingly locked behind $20-to-$50 monthly paywalls, mandatory account creation, and invasive identity harvesting.
• Treating humans as obsolete: Corporate marketing routinely frames human workers as inefficient bottlenecks, telling writers, artists, and engineers that their skills are merely “commoditized legacy baggage.”

It is easy to criticize that playbook. But when we started building Merciful, we wanted to ask a practical question: What happens if you design an AI tool from behind the blindfold? What does software look like when it is built for the person waiting at the other end of the table—the person who has to take whatever slice is left?

That single question is the engineering compass behind every decision we make. Here is why we do what we do—and why we believe it is fundamentally better for you.

1. Why We Built a Tool, Not a Companion (And Why That Protects Your Mind)

The biggest trend in consumer AI over the past two years has been the push toward virtual companions: chatbots that simulate emotional reciprocity, use pet names, remember intimate emotional details, and pose as synthetic therapists or best friends.

From a venture capital perspective, the motive is obvious: emotional attachment is the ultimate retention hook. If an algorithm convinces you it cares about you, you will never cancel your subscription.

Why this fails the cake-cutting test: If you were the vulnerable person behind the blindfold—someone going through a rough breakup, struggling with loneliness, or feeling overwhelmed—the last thing you need is a commercial algorithm engineered to exploit your biological attachment system. A neural network does not have feelings, moral agency, or genuine empathy; it is predicting statistical text continuations. When people substitute simulated relationships for real human connection, they retreat from the very friendships, families, and communities where real healing and mutual accountability actually happen.

Why we do what we do: At Merciful, our foundational rule is simple: AI is a tool, not a friend or a therapist. We design Merciful like an unpretentious utility—a digital scratchpad, a telescope, or a notebook that talks back.

You open it to untangle a confusing dilemma, brainstorm a story, pressure-test a business decision, or draft a difficult email without fear of judgment. It gives you clear, neutral, intelligent feedback. And then—most importantly—you close the tab and bring those thoughts to real people in your life. It respects your emotional boundaries, protects your independence, and keeps you in the driver’s seat.

2. Why We Removed the Toll Booths on Basic Thought (And Built for Every Device)

High-grade reasoning models are increasingly being locked behind walled gardens: mandatory sign-ups, phone number verification, and steep recurring subscription fees.

On top of financial barriers, there is an invisible hardware barrier. Most modern AI interfaces assume you are sitting in front of a high-end laptop or the latest smartphone. They are built as heavy, multi-megabyte applications demanding modern JavaScript runtimes, high memory footprints, and bleeding-edge browsers—quietly locking out older computers, low-power devices, legacy operating systems, and basic school or library terminals.

Why this fails the cake-cutting test: In 2026, AI is no longer a party trick; it is foundational cognitive infrastructure, like electricity, a calculator, or a public library. If you woke up behind the blindfold as a student writing their first research paper, a job seeker polishing their resume on an old hand-me-down machine, or a parent trying to understand a confusing medical bill, why should the ability to think through complex problems be an exclusive luxury good? True fairness is not just about having a free tier; it is about whether the software actually runs on the machine in front of you.

Why we do what we do: You should never have to surrender your personal phone number, hand over your private data, punch in a credit card, or own a thousand-dollar device just to sit down and think.

That is why Merciful is engineered without mandatory sign-ups, tracking cookies, or paywalls blocking everyday use—and why we deliberately prioritize wide, lightweight compatibility. Instead of assuming a modern, app-like browsing environment, we keep our client lightweight, resilient, and accessible across older machines and legacy device types. You open the page, type what is on your mind, and begin. Just like a clean desk at a public library, anyone can walk in, sit down, and start working.

3. Why Software Should Build Your Confidence, Not Make You Feel Obsolete

A lot of frontier tech rhetoric talks down to human beings. Industry manifestos frequently proclaim that human learning is pointless, that dedicated craft has been rendered worthless by compute clusters, and that people are simply waiting around to be automated out of relevance.

Why this fails the cake-cutting test: In Rawls’ philosophy, the single most critical primary good in any society is what he called the social bases of self-respect. Self-respect is the secure belief that your life, your skills, and your moral agency are meaningful, and that you have the standing to contribute to the world as an equal. Any technology that humiliates its users, treats them as passive eyeballs to be monetized, or tells them they are obsolete violates that foundation.

Why we do what we do: We believe technology should serve as what Steve Jobs famously called a “bicycle for the mind.” A bicycle does not pedal itself; it takes the human energy you already have and amplifies it so you can travel ten times farther.

Merciful is not designed to replace your voice, think for you, or push you out of the creative loop. It is an honest sounding board that helps you articulate your own ideas, spot gaps in your own arguments, and build confidence in your own work. The human remains the protagonist; the machine remains the instrument.

Designing from the Other Side of the Knife

When you build technology with the cake-cutting rule in mind, you stop asking: “How do we extract another three minutes of daily screen time from this user?”

Instead, you ask: “Does this software leave the person clearer, calmer, and more capable than they were when they opened the page?”

That is why Merciful is quiet, fast, and unpretentious. We don’t want you living inside our interface. We want you to solve what is in front of you, feel good about your work, and get back to your life.

— The Merciful Team