The Area Under the Curve: How AI Expands Human Work Capacity

The overwhelming volume of work that needs to be done (click for full size)

Every minute, millions of security events flow through corporate networks. Thousands of telescopes capture asteroids that could threaten Earth. Medical researchers analyze countless genetic sequences looking for disease patterns. And millions of hours of video are captured—many of which include crimes being committed.

But nobody’s paying attention.

Not because we don’t care, but because there’s just too many things to watch. To do. To monitor. To take action on.

The Universe of Intelligence Tasks

Our modern world runs on what I call Intelligence Tasks—work that requires human judgment, pattern recognition, and decision-making. These aren’t things you can solve with simple automation or basic programming. They require actual intelligence.

Here’s just a small sample of Intelligence Tasks happening (or not happening) right now:

Security & Safety

• monitor_security_cameras – Watch for suspicious activity
• analyze_network_traffic – Detect cyber intrusions
• review_access_logs – Find unauthorized access attempts
• investigate_fraud_claims – Determine if claims are legitimate
• track_space_debris – Monitor objects that could hit satellites

Medical & Health

• analyze_xrays – Look for abnormalities
• check_moles – Identify potential skin cancer
• review_patient_history – Find patterns in symptoms
• monitor_vital_signs – Detect concerning changes
• analyze_genetic_data – Identify disease markers

Business Operations

• review_contracts – Check for issues and risks
• process_insurance_claims – Determine validity and payout
• analyze_customer_feedback – Extract insights and trends
• quality_inspection – Find defects in products
• evaluate_loan_applications – Assess creditworthiness

Research & Analysis

• analyze_satellite_imagery – Track military movements
• review_scientific_papers – Extract key findings
• monitor_social_media – Detect emerging threats
• analyze_financial_data – Find trading opportunities
• investigate_corruption – Uncover illegal activities

The list goes on endlessly. Every industry, every field, every aspect of modern life generates Intelligence Tasks faster than we can possibly handle them.

What Makes Something an Intelligence Task?

Let’s look at a concrete example to understand what we’re talking about. Take CutePup, a company that curates cute dog photos for their website. Their process might seem simple, but it perfectly illustrates the concept:

Even “simple” tasks require human intelligence when you can’t code rules for them

This workflow has three Intelligence Tasks:

1. Is it a dog? – Requires visual pattern recognition
2. Is it cute? – Requires subjective judgment
3. What breed is it? – Requires specialized knowledge

You can’t write traditional code to do these things. You need intelligence—either human or artificial.

Now imagine Chris, who works at CutePup. He sits at his desk all day looking at uploaded photos and clicking “Yes” or “No” on whether they contain dogs. His colleague Carol determines if the dogs are cute. Amir identifies the breeds.

CutePup employs 48,912 people just to process their daily photo uploads. Nearly 50,000 humans doing work that requires intelligence but is relatively simple.

The Complexity Spectrum

Not all Intelligence Tasks are created equal. Let’s look at a more complex example: ClaimRight Insurance.

ClaimRight processes insurance claims for products that wear out prematurely. Their workflow shows how Intelligence Tasks can require significant expertise:

Multiple Intelligence Tasks requiring experience and judgment

Their pipeline includes:

1. Analyzing 50 photos per claim
2. Determining coverage eligibility
3. Reviewing video testimony
4. Verifying identity through face/voice
5. Matching items across media
6. Distinguishing wear from abuse
7. Approving or denying payout

Meet Kira, one of their top performers. With 25 years of experience, she processes 29 cases per day with 89% accuracy—exceptional by human standards. But ClaimRight needs 349,219 employees to handle their claim volume.

The jump in complexity from CutePup to ClaimRight is significant, but let’s go even further.

When Intelligence Tasks Require Extreme Expertise

Some Intelligence Tasks demand not just intelligence, but deep expertise built over decades. Consider Overseer, a military intelligence company analyzing satellite imagery:

High-stakes analysis requiring years of specialized training

Their daily workflow:

1. Process 28,452 new satellite images
2. Compare with previous day’s imagery
3. Identify all objects and changes
4. Assess military significance
5. Correlate with other intelligence
6. Write targeted reports for different agencies

Kevin, one of their star analysts, can produce 9 complete intelligence reports per week. That’s considered exceptional—he’s one of the few who can work across multiple parts of the pipeline. But even with 712,309 employees, Overseer can only analyze a fraction of what needs attention.

Or take BadSpot, a medical service checking for dangerous moles:

Life-or-death decisions requiring medical expertise

Every person working this pipeline must be:

• A licensed medical doctor (8+ years training)
• Dermatology specialized (3-4 additional years)
• Experienced in pattern recognition
• Capable of making life-or-death decisions

The result? Millions of people with suspicious moles never get them checked by a qualified professional. There simply aren’t enough doctors.

Visualizing the Work That Needs to Be Done

Now that we understand Intelligence Tasks, let’s visualize the scale of the problem. This chart represents all the Intelligence Tasks that exist in our world:

The total Intelligence Tasks that exist in our world

The x-axis represents volume—how many tasks need to be done. Think millions of insurance claims, billions of security events, trillions of financial transactions.

The y-axis represents difficulty—the expertise and intelligence required. From “is this a dog?” at the bottom to “diagnose this rare disease” or “assess this military threat” at the top.

The area under the curve? That’s everything that needs intelligent analysis to keep our civilization running smoothly.

The Harsh Reality of Human Capacity

Now let’s overlay what humans can actually accomplish:

We’re only covering a tiny corner of what needs to be done

That tiny blue area represents the sum total of human capacity. Every doctor, every analyst, every investigator, every expert on Earth working at full capacity.

Remember our examples:

• Kira processes 29 insurance cases per day (exceptional performance)
• Kevin produces 9 intelligence reports per week (genius level)
• A radiologist might read 100-500 scans per day (with fatigue)
• Chris reviews maybe 2000 dog photos per day (simple task)

Even with billions of humans, we can only handle:

• Small volumes of work (relative to what exists)
• Lower difficulty tasks (most of the time)
• A tiny fraction of what needs attention

Enter AI: Expanding Our Capacity

This is where AI fundamentally changes the equation. AI doesn’t just help us work faster—it expands both axes of our capacity:

AI dramatically expands our collective ability to handle Intelligence Tasks

Volume Expansion

Where Kira processes 29 insurance cases daily, an AI system could process 29,000. Where a security analyst reviews 100 alerts, AI can analyze millions. This isn’t just “working faster”—it’s operating at a fundamentally different scale.

Difficulty Expansion

AI can also tackle tasks requiring extreme expertise:

• Medical diagnosis requiring 12+ years of training
• Military analysis needing decades of experience
• Pattern detection too subtle for human perception
• Correlation across massive, disparate datasets

The KISAC Framework: Measuring Intelligence Task Performance

To understand why AI can expand both dimensions so dramatically, consider what makes someone good at Intelligence Tasks:

• Knowledge — All the information, training, and experience
• Intelligence — Ability to find patterns and generate insights
• Speed — How many tasks completed per time period
• Accuracy — Correctness and error rates
• Cost — Total expense to employ and maintain

Let’s compare:

Metric	Top Human Performance	AI Performance
Knowledge	Decades of experience, thousands of cases	All human knowledge, millions of examples
Intelligence	IQ ~180 maximum, degrades with fatigue	Approaching human level, improving rapidly
Speed	29 insurance cases/day (Kira)	29,000+ cases/day
Accuracy	89% on insurance fraud (exceptional)	93%+ and improving
Cost	$137,200/year salary + benefits	$3,500/year compute costs

The performance gap is already massive and growing

What This Means for Society

The implications are profound:

1. Most Intelligence Tasks aren’t being done at all – There’s no human available
2. AI can fill the gap – Not by replacing humans, but by doing work that was never getting done
3. Both volume and difficulty expand – AI handles more tasks AND harder tasks
4. The focus should be on coverage – How do we ensure important work gets done?

Think about all the:

• Fraud that goes uninvestigated
• Diseases that go undiagnosed
• Security threats that go undetected
• Research that never happens
• Corruption that goes uncovered
• Insights that remain hidden

A New Model for Understanding AI’s Role

Understanding work as “area under the curve”—combining both volume and difficulty—gives us a clearer picture of AI’s true impact. It’s not about replacement. It’s about expansion.

Every Intelligence Task that goes undone has real consequences. Every uninvestigated crime, every undiagnosed disease, every undetected threat represents a failure not of effort, but of capacity.

AI offers us a way to dramatically expand that capacity, to illuminate the dark corners of work that we’ve never been able to reach.

Summary

• Our world generates vastly more Intelligence Tasks than humans can possibly handle
• These tasks span from simple (is it a dog?) to complex (diagnose this disease)
• Human capacity is a tiny corner of what needs to be done—limited in both volume and difficulty
• Real organizations need massive human workforces just to handle fractions of their Intelligence Tasks
• AI expands our capacity on both axes: handling more volume AND higher difficulty
• The real opportunity isn’t replacing human work—it’s finally doing the critical work that’s never been done