Beyond Bias Is Coming. Here's Why It Matters.
Beyond Bias Is Coming. Here’s Why It Matters.
This is the post I have been building toward for twenty one articles.
Over the past five months, I have taken you through a landscape that most people only encounter as headlines. An eleven year old girl who died of starvation after her family’s ration card was cancelled for not being linked to Aadhaar. A hiring algorithm that taught itself to reject women. Credit scoring systems that use your PIN code as a proxy for your caste. Facial recognition technology that fails on dark skinned faces at rates a hundred times higher than on light skinned ones. Deepfakes of dead politicians endorsing living candidates in the largest election the world has ever seen. Five hundred million Indians invisible to the AI systems that are increasingly deciding who gets a loan, who gets a job, who gets government benefits, and who gets nothing.
I showed you these stories not to frighten you, but to make you see what I see every day from the engineering floor: the gap between what AI can do and what it should do is vast, widening, and unacceptably dangerous.
Today, I am going to tell you what I have been doing about it.
The Book
Beyond Bias: The Four Way Test for Ethical AI is a book I have spent two years writing. It arrives at a moment when India has over eight hundred million internet users coming under a comprehensive privacy law for the first time, when the country has deliberately chosen a light touch approach to AI governance with zero enforcement actions taken, and when the question of who is accountable when AI makes decisions about your life remains, in most jurisdictions, unanswered.
The book is ten chapters, approximately sixty to seventy five thousand words, and it covers the complete landscape of AI ethics as it stands in 2026, with a framework that makes it actionable.
Here is what it covers and why each piece matters.
What The Book Covers
Chapter 1: The Machine That Learns. Before we can talk about what AI should do, we need to understand what it actually is. Not the hype. Not the fear. The reality. I strip away the jargon and explain how AI works using an analogy I have tested on everyone from my mother to my CTO: the warehouse analogy. If you can understand a warehouse full of documents and a person hired to read them and answer questions, you can understand how a large language model works. This chapter also maps the most important shift happening right now: AI moving from tools that assist to agents that act.
Chapter 2: Where Bias Lives. Bias is not a single thing. It enters AI systems through the data they are trained on, the algorithms that process that data, and the contexts in which they are deployed. This chapter tracks the lifecycle of bias from collection to consequence, with deep attention to biases unique to India: linguistic bias in language models that think in English, caste encoded as proxy variables in lending algorithms, the digital divide between urban and rural India that means half the country is invisible to the systems making decisions about their lives.
Chapter 3: The Black Box Problem. When an AI system denies your loan application, rejects your resume, or recommends a medical treatment, can anyone explain why? This chapter confronts the opacity of AI decision making and makes the case that transparency is not a luxury. It is the prerequisite for every other ethical principle. If you cannot see inside the system, you cannot audit it for bias, verify its fairness, or hold anyone accountable for its failures.
Chapter 4: Fair By Design. Fairness sounds simple until you try to build it into a system. This chapter explains the mathematically proven impossibility theorem, the fact that you cannot satisfy all fairness criteria simultaneously, and then shows you how to navigate that impossibility. Using India’s own reservation system as a central analogy, it demonstrates that fairness was never about finding a perfect solution. It was always about making defensible tradeoffs and being honest about them.
Chapter 5: Your Data, Their Model. Your data is training AI models you have never heard of, and the consent you gave was designed for a world that no longer exists. This chapter examines the privacy crisis in full, from Cambridge Analytica’s legacy to India’s Digital Personal Data Protection Act and the gaps that remain. It confronts the question that no one has adequately answered: what does meaningful consent look like when the thing you are consenting to is incomprehensible?
Chapter 6: When AI Acts. This is the chapter that most AI ethics books do not have, because the problem is too new. Agentic AI, systems that do not just recommend but act, autonomously, without waiting for human permission, is the frontier. This chapter maps the accountability crisis that emerges when AI agents negotiate contracts, execute trades, screen candidates, and interact with other agents, all without a human in the loop. It includes the Autonomy Threshold Framework, a practical tool for deciding when a human must intervene.
Chapter 7: The Regulatory Landscape. The EU chose prescription. The US chose fragmentation. China chose targeted intervention. India chose to watch, build, and wait. This chapter maps the global regulatory experiment in real time and examines India’s approach on its own terms, including the seven sutras, the AI Ethics and Accountability Bill, and what zero enforcement actions tells us about the gap between policy intention and policy reality.
Chapter 8: The Hidden Costs. Training a single large AI model like GPT-4 consumed roughly 50,000 megawatt hours of electricity, enough to power tens of thousands of Indian households for a year. Entry level hiring has dropped dramatically as AI handles tasks that used to be someone’s first job. Data labelling, the invisible labour behind every “intelligent” system, is performed by workers in the Global South under conditions that would make headlines if the products were physical. This chapter exposes the costs that never appear on a balance sheet.
Chapter 9: The Four Way Test in Practice. This is the operational chapter. Everything the book has built, every principle, every case study, every cautionary tale, comes together here as a working toolkit. Role specific implementation guides for developers, product managers, business leaders, and policymakers. A complete AI ethics audit walkthrough. Templates for setting up an ethics review board, a pre deployment checklist, and an incident response plan for when things go wrong.
Chapter 10: What Comes Next. The future chapter. Where the trajectories mapped throughout the book converge. Not prediction as prophecy, but prediction as logical extrapolation from the conditions we can see today. Where AI capabilities are heading. Where regulation is heading. Where India sits in the global picture. And how the Four Way Test evolves as the technology it governs becomes more powerful.
Who It Is For
I wrote this book for four people.
The product manager who has to decide, this Thursday, whether to deploy a model that is ninety two percent accurate overall but seventy eight percent accurate for rural users. She needs a framework for making that decision and a way to document her reasoning that will hold up to scrutiny.
The developer who knows, in his gut, that the training data is skewed, but does not have the language or the organisational authority to slow the release. He needs a checklist that gives him that language and that authority.
The business leader who is signing procurement contracts for AI systems without understanding what questions to ask about bias, transparency, or accountability. She needs a set of criteria that separates vendors who take ethics seriously from vendors who put “responsible AI” on a slide deck and move on.
The policymaker who is drafting India’s next AI guideline and needs to understand not just what other countries are doing, but whether those approaches actually work, and what India’s specific context demands.
If you are any of these people, this book was written for you. If you are none of these people but you use AI tools, are affected by AI decisions, or care about how these systems shape the world your children will inherit, this book was written for you too.
What Makes It Different
Three things.
First, it is practical. Every chapter ends with tools you can use. Not principles to contemplate. Tools to deploy. Checklists, audit frameworks, decision templates, role specific guides. You should be able to use this book on Monday morning.
Second, it starts from India. The case studies are Indian. The regulatory analysis centres on Indian law. The fairness discussion grapples with caste, linguistic diversity, and the urban rural digital divide. This is not a Western framework with an India footnote. It is a book written from the engineering floor in India, for the reality of building and deploying AI in India.
Third, it has a framework that fits in your head. The Four Way Test. Four questions. Truth, fairness, privacy, accountability. Simple enough to remember, rigorous enough to apply, flexible enough to scale from a startup’s product review to a government ministry’s procurement decision. Built from values my grandfather taught me and battle tested against every AI failure case I could find.
The Trilogy
Beyond Bias is the second book in what I think of as a trilogy.
The first, Talking to AI, was about understanding AI as a user. How to communicate with AI tools effectively, how to get better results, how to use these systems as the powerful tools they are.
Beyond Bias is about understanding AI as a builder, a deployer, a regulator, a citizen. It is the ethics book. The one that asks: now that we can build these things, what rules should govern them?
The third, Universe of Algorithms, is the bigger picture. How algorithms shape the systems we live inside, from social media to financial markets to democratic processes. That one is in progress.
Together, the three books cover the full arc: how to use AI, how to govern AI, and how AI governs us. Beyond Bias is the critical middle piece.
Why Now
There is a window. I described it in the book’s foreword, and I will describe it again here because it is important.
We are in the window between voluntary guidance and inevitable regulation. Between “we trust companies to do the right thing” and “we will force companies to do the right thing.” India’s AI Governance Guidelines are voluntary. The Data Protection Board has not brought a single case. The AI Ethics and Accountability Bill is a Private Member’s Bill with the statistical likelihood of passage that all Private Members’ Bills carry, which is to say, almost none.
But the trajectory is clear. Every jurisdiction that starts with voluntary guidelines eventually moves toward binding regulation. The EU took that journey. India will take it too. The question is not whether regulation will come, but whether the regulation that comes will be reasonable or punitive.
What businesses and builders do in this window matters enormously. If the industry demonstrates that it can self govern, that it can apply frameworks like the Four Way Test proactively and credibly, the regulation that follows will be collaborative. If the industry does what the Indian technology industry has historically done in other domains, which is to treat self regulation as a slogan rather than a practice, the regulation that follows will be harsh.
This is the window where a simple, practical ethical framework matters most. Before it is mandated. While it is still a choice.
What I Am Asking You To Do
If any of this resonates, I am asking one thing.
Register to be notified when the book launches. That is it. I am not asking you to buy anything today. I am asking you to raise your hand and say: I want to know when this is ready.
Because the people who raise their hand now are the people I am writing for. The ones who are already in the room when the decisions are made. The ones who already feel the gap between knowing and doing. The ones who want something better than good intentions and a backlog ticket labelled “ethics audit — deferred.”
This book is that something better. Four questions. Every AI decision. Monday morning.
I’m have written a book about exactly this how AI and automated systems make decisions about your life, where accountability disappears, and what we can do about it. If you want to know more about this book or order a copy, you can do it here: https://akshaywalimbe.com/beyond-bias/