Cambridge AI Ethicist’s Book ‘What If We Got AI Right?’ Falls Short

A new book by Cambridge researcher Eleanor Drage, 'What If We Got AI Right?', argues that the public must look past Silicon Valley's narratives to see the human labor and environmental costs of AI, but critics say it fails to offer scalable, specific remedies. Drage, an assistant research professor at the University of Cambridge's Leverhulme Centre for the Future of Intelligence, highlights community-driven projects like Mumkin and Kuini while dismissing unconscious bias training, yet the book does not engage with AI's widely praised applications in drug discovery or early disease detection.

July 27, 2026, Inside AI — A new book by Cambridge researcher Eleanor Drage aims to reframe how society thinks about artificial intelligence, but critics say it stumbles when moving from diagnosis to prescription. What If We Got AI Right? argues that the public must see past Silicon Valley’s utopian and apocalyptic narratives to recognize the human labor and environmental costs embedded in AI systems. Drage, an assistant research professor at the University of Cambridge’s Leverhulme Centre for the Future of Intelligence, insists that demystifying AI is the first step toward reclaiming power from tech giants. She points out that “the cloud” is simply someone else’s computer, and that an AI “hallucination” is a system error rather than a sign of sentience. Her goal is to shift conversations away from existential risk and toward tangible safety measures, such as data rights and model regulation. The book arrives at a moment when global AI governance is splintering. The EU’s AI Act has set binding rules for high-risk systems, while the U.S. relies largely on voluntary commitments. Drage’s focus on labor conditions in the AI supply chain echoes findings from a 2023 paper on data labor exploitation https://arxiv.org/abs/2303.10130 that documented how workers in Kenya and Venezuela label toxic content for less than $2 per hour. Yet her own proposed remedies have drawn fire for lacking scale and specificity. Drage highlights community-driven projects like Mumkin, an app designed to facilitate conversations about female genital mutilation in India, and Kuini, a chatbot helping Māori women quit smoking. But reviewers question whether these tools deliver value that simpler, non-AI interventions could not. The book does not engage with AI’s widely praised applications in drug discovery—such as DeepMind’s AlphaFold, which has predicted protein structures for over 200 million proteins—or in early disease detection, leaving its central question unanswered. Labor Shifts and the Unconscious Bias Debate One of Drage’s more contentious claims is that “work is simply shifting, from artists to data labellers and from marketeers to engineers.” Critics note that this framing glosses over the precarity of data labelers, who often earn poverty wages and lack labor protections. A 2022 study on AI supply chains https://arxiv.org/abs/2204.05433 found that the global market for data annotation is projected to reach $13.7 billion by 2030, yet workers remain invisible in policy debates. Drage also dismisses unconscious bias training as “merely another way of taking the perspective of the perpetrator rather than the recipient of harm.” This stance has drawn pushback from psychologists who argue that implicit bias research, while imperfect, provides a framework for understanding systemic discrimination. Her rejection of intent as irrelevant may oversimplify a debate where both impact and motivation matter. Utopia as Process, Not Destination Drage describes a compliance toolkit she co-designed for AI companies as a “cornerstone of utopianism,” redefining utopia as the process of working together toward fairness. Yet the book stops short of detailing how AI itself—beyond ethics checklists—can drive that process. It does not mention, for example, how federated learning could keep personal data local while improving models, or how algorithmic auditing tools are being deployed to detect bias in hiring. The book’s strength lies in its philosophical rigor around defining intelligence and its critique of tech’s apocalyptic framing, which Drage argues distracts from climate change and inequality. But as a roadmap for getting AI right, it leaves readers with more questions than answers. For a field desperate for actionable visions, that may be the most disappointing hallucination of all.