## Predictive Algorithms and Machine Learning

What if a computer could predict that you were about to commit a crime, before you had even thought about it? It is the central premise of *Minority Report*, and while the film uses psychic "precogs" rather than algorithms, the idea of using data and computation to predict human behavior is very much a reality. The question is not whether we can do it, but how well it works and what happens when it gets things wrong.

### What Are Predictive Algorithms?

Predictive algorithms are software systems that analyze large datasets to identify patterns and make forecasts about future events or behaviors. They are a core application of machine learning, a branch of artificial intelligence where systems improve their performance by learning from data rather than being explicitly programmed.

These algorithms are everywhere. They recommend what you watch on streaming services, determine what appears in your social media feed, assess your creditworthiness, set insurance premiums, and flag potentially fraudulent transactions. In the criminal justice system, predictive algorithms are used for everything from identifying crime hotspots to assessing the likelihood that a defendant will reoffend.

The power of these systems comes from their ability to process vast amounts of data and detect patterns that would be invisible to humans. The danger comes from the same source: patterns extracted from historical data can encode and perpetuate existing biases, and the opacity of many machine learning systems makes it difficult to understand or challenge the basis for their predictions.

### How the Book Explores It

*Films from the Future* (Chapter 4) uses *Minority Report* to explore predictive algorithms and their implications for justice. In Steven Spielberg's 2002 film, based on a Philip K. Dick story, a special police unit called Precrime uses three psychics to foresee murders before they happen. Suspects are arrested and imprisoned for crimes they have not yet committed. The system appears to work flawlessly, until the unit's own chief discovers that he has been predicted to commit a murder he has no intention of carrying out.

The book draws a direct line from this fictional scenario to real-world predictive policing systems. It examines how algorithms trained on historical crime data can reinforce the same patterns of racial and socioeconomic bias that are embedded in the data. A system trained on arrest records from neighborhoods that are already heavily policed will predict more crime in those neighborhoods, leading to more policing, more arrests, and more data that confirms the original prediction. The result is a feedback loop that looks objective but encodes systemic bias.

The book also explores the fundamental problem of false positives: predictions that someone will do something harmful when they would not have. In *Minority Report*, the discovery that the precog system produces conflicting predictions, known as "minority reports," reveals that the entire system rests on a lie. In the real world, every predictive system has a false positive rate, and the consequences of acting on false predictions in the criminal justice system can be devastating.

### Where Things Stand Today

Predictive algorithms are being deployed in criminal justice, hiring, lending, healthcare, education, and dozens of other domains. In some cases, they have improved efficiency and outcomes. In others, they have produced discriminatory results that have prompted legal challenges and public backlash.

The field of AI fairness has grown substantially in response to these concerns. Researchers are developing techniques to detect and mitigate bias in algorithmic systems, and there is growing regulatory interest in requiring transparency and accountability for automated decision-making. But the fundamental tension between the power of these systems and the opacity of their decision-making remains unresolved.

### Why It Matters

Predictive algorithms matter because they increasingly determine who gets a job, who gets a loan, who gets surveilled, and who goes to prison. These are decisions that shape lives, and when they are delegated to opaque computational systems, the normal mechanisms of accountability, appeal, and due process can break down.

The lesson from *Minority Report*, and from the book, is that prediction is not the same as certainty. Every predictive system makes mistakes, and those mistakes fall disproportionately on people who are already marginalized. Building algorithmic systems that are fair, transparent, and accountable is not just a technical challenge; it is a moral one.

### Explore Further

- [Artificial Intelligence](https://spoileralert.wtf/md-files/est_artificial_intelligence.md) — the broader field that predictive algorithms are part of
- [Ubiquitous Surveillance and Big Data](https://spoileralert.wtf/md-files/est_surveillance.md) — the data infrastructure that feeds predictive systems
- [Surveillance, Privacy, and Control](https://spoileralert.wtf/md-files/rei_surveillance_privacy_control.md) — the ethical implications of algorithmic monitoring
- [Automation and Robotics](https://spoileralert.wtf/md-files/est_automation.md) — another domain where algorithms replace human judgment

## Further Reading

- [Minority Report: Predicting Criminal Behavior — Moviegoer's Guide to the Future (Future of Being Human)](https://www.futureofbeinghuman.com/p/minority-report-predicting-criminal) — Andrew Maynard uses *Minority Report* to explore predictive policing, algorithmic bias, and the fundamental problem of acting on predictions that may be wrong, drawing a direct line from the film's fiction to real-world criminal justice algorithms.
- [Learning to Live with Agentic Social AI — Andrew Maynard (Future of Being Human, 2024)](https://www.futureofbeinghuman.com/p/learning-to-live-with-agental-social-ai) — Maynard explores how AI systems could exploit human cognitive biases and gain agency through their ability to manipulate human behavior — extending the *Minority Report* chapter's concerns into the era of agentic AI systems that can actively influence the people they model.
- [EU AI Act, Article 5: Prohibited AI Practices (Enforceable February 2025)](https://artificialintelligenceact.eu/article/5/) — The primary legal text banning AI-based predictive policing that profiles individuals based on personality traits or personal characteristics, with penalties up to €35 million or 7% of global turnover. The first major regulatory framework to classify predictive policing as an "unacceptable risk" — essential context for the global divergence between the EU ban and continued US deployment.
- [Algorithmic Fairness: Challenges to Building an Effective Regulatory Regime — Demirchyan, *Frontiers in Artificial Intelligence* (August 2025)](https://www.frontiersin.org/journals/artificial-intelligence/articles/10.3389/frai.2025.1637134/full) — A peer-reviewed examination of why legislating algorithmic fairness is so difficult: experts disagree on what fairness means, statistical parity can produce counterintuitive results, and cross-jurisdictional consistency remains elusive. Argues that effective regulation may require an FDA-style approval process for high-stakes algorithms.
- [Algorithmic Fairness in Predictive Policing — *AI and Ethics* (Springer, 2024)](https://link.springer.com/article/10.1007/s43681-024-00541-3) — Using the Chicago Strategic Subject List dataset, this study found that all 118,146 individuals under 30 were categorized as high risk — demonstrating how feedback loops and narrow fairness definitions can create systematic discrimination that standard bias audits miss, while revealing that existing fairness strategies focus primarily on racial bias while overlooking age, gender, and socioeconomic dimensions.
- [NAACP Artificial Intelligence in Predictive Policing Issue Brief (2025)](https://naacp.org/resources/artificial-intelligence-predictive-policing-issue-brief) — A major civil rights policy document calling on state legislators to regulate predictive policing tools, citing mounting evidence that these systems increase racial bias and erode public trust in Black communities. Includes five concrete policy recommendations covering oversight, transparency, and community engagement.
- [Palantir's All-Seeing Eye: Domestic Surveillance and the Price of Security (SETA Foundation, 2025)](https://www.setav.org/en/palantirs-all-seeing-eye-domestic-surveillance-and-the-price-of-security) — Named after Tolkien's seeing-stones, Palantir Technologies now underpins immigration enforcement, predictive policing, and inter-agency intelligence-sharing across the US, UK, and Europe — the Precrime system of *Minority Report* made real, piece by piece.
- [New Orleans Program Offers Lessons in Pitfalls of Predictive Policing (ACLU)](https://www.aclu.org/news/privacy-technology/new-orleans-program-offers-lessons-pitfalls-predictive-policing) — The ACLU's analysis of how Palantir's secret six-year partnership with the New Orleans Police Department used social media, criminal records, and gang affiliations to generate predictive scorecards — deployed without public knowledge, procurement process, or city council approval.