Book Review: ‘Prophecy,’ by Carissa Véliz

Why It Matters

Predictive AI underpins decisions in finance, law enforcement and personal data, so mischaracterizing it as neutral truth threatens liberty and market fairness. Véliz’s critique equips policymakers and businesses to address concrete risks before they become entrenched.

Key Takeaways

• Big Tech profits from predictive algorithms that shape consumer behavior
• Generative AI models predict plausible text, not factual truth
• Véliz warns focus on existential AI risk distracts from privacy threats
• Historical prophecy parallels modern data forecasting, raising freedom concerns

Pulse Analysis

The rise of predictive technology is not a novel phenomenon; societies have long relied on oracles, insurance actuarial tables, and statistical risk models to anticipate the future. Véliz places today’s AI systems on that continuum, showing how massive data collection has turned prediction into a commercial engine. By quantifying human behavior—whether a borrower will default or a citizen might reoffend—big‑tech firms monetize foresight, creating a feedback loop where the forecast itself shapes outcomes. This historical perspective underscores that the power to predict has always carried political weight, now amplified by global data networks.

Véliz’s core argument is that generative AI, especially large language models, are designed to be fortune tellers, not truth tellers. They stitch together statistically likely continuations of text without grounding in verified facts, leading to hallucinations that appear authoritative. When such systems inform hiring, policing or credit decisions, the veneer of objectivity masks underlying biases and privacy intrusions. The book warns that the public’s fascination with existential AI catastrophes—paper‑clip scenarios or killer robots—distracts from these immediate, tangible harms: erosion of personal autonomy, opaque decision‑making, and the commodification of intimate data.

For regulators and corporate leaders, Véliz’s insights suggest a pivot from speculative risk assessments to concrete governance frameworks. Transparency mandates, data‑minimalism, and algorithmic audits become essential tools to ensure predictions serve public interest rather than corporate profit. By reframing AI as a predictive oracle that must be held accountable, stakeholders can craft policies that protect freedom while still leveraging the legitimate benefits of foresight. The conversation shifts from fearing a distant singularity to safeguarding today’s democratic values against the subtle tyranny of data‑driven prophecy.