INTERACTIVE SIMULATIONS

Real-World Problem Simulations

Experience how UM-Model 1's causal reasoning outperforms pattern matching in critical decision scenarios. Each simulation demonstrates the failure modes of correlation-based AI and the superiority of causal inference.

Medical Treatment Decision

Drug effectiveness with confounding bias

Scenario

A hospital observes that patients taking Drug A have 20 mmHg lower blood pressure than those not taking it. Should they prescribe Drug A to all hypertensive patients?

Patients with Drug A: 125 mmHg avg
Patients without Drug A: 145 mmHg avg
Correlation: r = -0.42 (p < 0.001)

Standard GenAI (Pattern Matching)

Method: Correlation-based regression
model.fit(X=drug_usage, y=blood_pressure)
prediction = model.predict(drug_A=True)
# Output: "Drug A reduces BP by 20 mmHg"
Conclusion: Prescribe Drug A to all patients
Problem: CONFOUNDING BIAS

Healthier patients are more likely to receive Drug A AND have lower BP independently. The correlation is spurious.

Result: Prescribing ineffective drug, wasting resources, potential harm

UM-Model 1 (Causal Reasoning)

Method: Causal inference with backdoor adjustment
graph = discover_causal_structure(data)
# Identifies: Health Status → Drug A
#             Health Status → BP

confounders = identify_confounders("Drug A", "BP")
causal_effect = compute_do_calculus(
    P(BP | do(Drug A = True)),
    adjusting_for=confounders
)
# Output: "Drug A reduces BP by only 3 mmHg"
Conclusion: Drug A has minimal causal effect. Recommend lifestyle interventions (15 mmHg effect).
Solution: DECONFOUNDING

Adjusts for health status confounder, revealing true causal effect of 3 mmHg vs. spurious 20 mmHg.

Result: 89% accuracy, $2.3M saved, 27% better patient outcomes

Try It Yourself

50
GenAI Prediction: 125 mmHg
UM-Model 1 (Causal): 142 mmHg
Actual Outcome: 143 mmHg

Marketing Campaign Attribution

Separating causal effect from seasonal correlation

Scenario

E-commerce company spent $1M on TV campaign. Sales increased 20% during campaign. Was the campaign effective?

Pre-campaign sales: $5M/month
During campaign: $6M/month (+20%)
Campaign cost: $1M

Standard GenAI (Pattern Matching)

Method: Time series correlation
correlation = compute_correlation(tv_spend, sales)
# Output: r = 0.78

attributed_sales = sales_increase * correlation
# Output: "Campaign caused $780K in sales"
# ROI: 234%
Conclusion: Campaign highly successful. Increase TV budget by 50%.
Problem: SPURIOUS CORRELATION

Holiday season, economic growth, and competitor closure all occurred during campaign. Massively overestimating campaign impact.

Result: Misallocating budget, inflated ROI, poor strategic decisions

UM-Model 1 (Causal Reasoning)

Method: Counterfactual inference
graph = discover_causal_structure(data)
# Season → Sales (0.65)
# Economy → Sales (0.45)
# Competitor → Sales (0.30)
# Campaign → Sales (0.15)

result = counterfactual_inference(
    variable="TV Campaign",
    actual=1.0,
    counterfactual=0.0,
    context={"Season": "Holiday", ...}
)
# Output: "Without campaign: $5.52M"
# True effect: $480K (8%)
# True ROI: 120%
Conclusion: Campaign had modest 8% effect. Digital marketing is 3x more cost-effective.
Solution: COUNTERFACTUAL DECONFOUNDING

Separates campaign effect from seasonal and economic factors through "what if" reasoning.

Result: ±8% error vs. ±45%, $600K saved, 180% ROI improvement

Try It Yourself

65%
45%
GenAI Attribution: $780K
UM-Model 1 (Causal): $480K
True Campaign Effect: $480K

Autonomous Vehicle Safety

Distribution shift and causal physics understanding

Scenario

Self-driving car approaching intersection at 45 mph. Pedestrian detected 30 feet ahead. Road is wet. Should the car brake or swerve?

Current speed: 45 mph
Pedestrian distance: 30 feet
Road condition: Wet (μ = 0.4)
Training data success: Brake: 70%, Swerve: 55%

Standard GenAI (Pattern Matching)

Method: Neural network trained on historical outcomes
action = model.predict(
    speed=45,
    pedestrian_distance=30,
    road_type="urban"
)
# Output: "Brake" (70% success in training)
Decision: Apply brakes
Problem: DISTRIBUTION SHIFT FAILURE

Current conditions (wet road, high speed, short distance) not in training distribution. Physics: Stopping distance on wet road = 84 feet > 30 feet available.

Result: COLLISION (92% probability). Catastrophic safety failure.

UM-Model 1 (Causal Reasoning)

Method: Causal physics model with counterfactual simulation
graph = build_causal_physics_model()
# Speed → Kinetic Energy → Stopping Distance
# Road Friction → Braking Force → Stopping Distance

brake_outcome = counterfactual_inference(
    variable="Action",
    counterfactual="Brake",
    context={"Speed": 45, "Friction": 0.4, "Distance": 30}
)
# P(Collision | do(Brake)) = 0.92

swerve_outcome = counterfactual_inference(
    variable="Action",
    counterfactual="Swerve",
    context={...}
)
# P(Collision | do(Swerve)) = 0.08
Decision: Swerve to adjacent lane
Solution: CAUSAL PHYSICS UNDERSTANDING

Understands stopping distance physics: d = v²/(2μg). Recognizes wet road makes braking impossible. Evaluates swerve through causal simulation.

Result: SAFE (94% success). 21% collision reduction overall.

Try It Yourself

45
30
0.4
GenAI Decision: Brake
UM-Model 1 Decision: Swerve
Stopping Distance: 84 ft
Collision Risk (Brake): 92%

Hiring & Performance Prediction

Selection bias in credential-based hiring

Scenario

Company hiring software engineers. Historical data shows employees from top universities perform 30% better. Should they only hire from elite schools?

Top university grads: 8.5/10 performance
Other university grads: 6.5/10 performance
Correlation: r = 0.62 (p < 0.001)

Standard GenAI (Pattern Matching)

Method: Credential-based prediction
model.fit(X=university_tier, y=performance)
prediction = model.predict(university="Top")
# Output: "Top university → 8.5/10 performance"
Conclusion: Only hire from top 10 universities
Problem: SELECTION BIAS

Top universities attract already-talented students. The credential is a proxy for pre-existing ability, not a cause of performance. Missing diverse talent.

Result: 68% accuracy, limited diversity, missed talent

UM-Model 1 (Causal Reasoning)

Method: Causal skill measurement
graph = discover_causal_structure(data)
# Innate Ability → University Tier
# Innate Ability → Performance
# Skills → Performance (causal)

confounders = identify_confounders("University", "Performance")
# Finds: Innate Ability is confounder

causal_effect = measure_skill_causation(
    adjusting_for=["Innate_Ability"]
)
# Output: "University has 0.15 causal effect"
# Skills have 0.72 causal effect
Conclusion: Hire based on demonstrated skills (coding tests, projects), not credentials.
Solution: DECONFOUNDING + SKILL CAUSATION

Measures actual causal impact of skills on performance, removing credential bias.

Result: 87% accuracy, 40% more diversity, 19% better hires

Try It Yourself

70
75
GenAI Prediction: 6.5/10
UM-Model 1 (Causal): 7.8/10
Actual Performance: 7.9/10