ITAI 1370 — AI History, Theory & Platforms

End-of-Year Portfolio — Cesar Zaldivar | May 2026

Module 12 — Week 12

Assignment A12: Understanding Predictive AI

Assignment TypeAssignment — Research Essay
TopicHow predictive AI anticipates human needs and automates decision-making
Date Submitted26 April 2026

What is Predictive AI?

While artificial intelligence has traditionally been defined by its ability to guess future states based on historical patterns, predictive AI is shifting toward anticipating human needs and automating decision-making. Ball (2024) argues that the focus is moving away from simple predictive analytics toward generative models capable of more sophisticated cognitive tasks (28). AI agents rely on a massive amount of collective data — as Mühlhoff (2025) highlights, AI agents learn by watching everyone, then use those collective patterns to make decisions about you as an individual (163).

Predictive AI works by treating human behavior as a series of mathematical coordinates. By processing this data, the system identifies recurring clusters then calculates the most probable next step based on those historical patterns. While predictive AI seems like it is thinking, the AI agent is actually calculating the statistical likelihood of your next move by comparing your current data point to the millions of others it has already processed.

Four-step predictive AI pipeline diagram

How Predictive AI Works — Step by Step

StepProcessExample
1Collect behavioral dataAll your past credit card transactions
2Identify recurring patternsYou always spend under $200 at grocery stores
3Calculate probability of next actionA $1,400 charge at an overseas retailer is statistically anomalous
4Predict and actFlag and block the transaction as likely fraud

Real-World Application: Credit Card Fraud Detection

Predictive AI has modernized credit card security by replacing rule-based systems with a dynamic behavioral fingerprint for every user. By analyzing vast amounts of historical data, the AI establishes a baseline of legitimate activity unique to the individual. When a new transaction occurs, the system evaluates it against that baseline. This allows financial institutions to proactively block suspicious charges while remaining less disruptive to the cardholder’s actual spending habits.

AI in Fiction vs. Reality: HAL 9000

In Stanley Kubrick’s 2001: A Space Odyssey, predictive AI is embodied by the HAL 9000, an onboard supercomputer that monitors crew behavior and anticipates mechanical failures. HAL’s portrayal of predictive maintenance is realistic — it mirrors how AI analyzes data to prevent breakdowns. The film becomes unrealistic when HAL’s logic predicts human intent and leads to preemptive and lethal action. HAL’s autonomous moral reasoning portrays an AI agent that is far more self-aware and strategic than anything that currently exists.

HAL 9000 FeatureRealistic?Explanation
Predictive maintenance of ship systems✓ RealisticMirrors modern industrial AI monitoring
Monitoring crew behavioral patterns✓ RealisticSimilar to workplace productivity AI
Predicting and acting on human intent✗ UnrealisticRequires moral reasoning AI cannot currently perform
Autonomous lethal decision-making✗ UnrealisticNo current AI operates with this level of agency

Creative Application: “Orbit”

As a creative component, I developed a storyline for a comedy called Orbit, featuring a smart-home assistant that gets too good at predicting its owner’s needs. Orbit starts solving life’s minor inconveniences: canceling a blind date because it predicts zero chemistry, ordering a large pizza when it detects a bad day at work, and muting your phone because it knows you’re not emotionally ready to talk to your mother yet. The story explores the tension between convenience and autonomy — what happens when your gadgets already have your life figured out?

What I Learned

Predictive AI is fundamentally about statistical pattern matching at massive scale, not genuine understanding or intuition. The credit card fraud detection example showed me how this technology already protects millions of people every day without them being aware of it.

The HAL 9000 analysis was a valuable exercise in separating realistic AI capabilities from science fiction. It reinforced that the danger of AI is not malevolent consciousness — it is the unintended consequences of optimization without ethical constraint. The Orbit story was a fun way to explore the social implications of predictive AI that knows us too well, raising the question: at what point does helpful prediction become unwanted control?

Citations

Ball, Matthew. “The Shift from Predictive to Generative AI.” 2024, p. 28.

Mühlhoff, Rainer. “Predictive Privacy.” AI and Society, 2025, p. 163.