Can AI Be Used as a Data Analyst?

An Experiment Using Pickup Trucks, Highway Behavior, and Generative AI

While developing my data analysis skills through real-world datasets and business scenarios, I wanted to answer a simple question:

Can AI help uncover meaningful insights from a problem even when a structured dataset doesn't exist?

To test this idea, I chose a topic that I've personally observed for years:

Why do pickup trucks sometimes appear to drive more aggressively on highways than other vehicles?

Instead of gathering a traditional dataset, I used ChatGPT as a research assistant to explore the question using publicly available information, studies, transportation research, psychology, and vehicle market trends.

The goal was not to prove that pickup truck drivers are more aggressive.

The goal was to determine whether AI could help generate useful hypotheses and identify patterns worth investigating further.

The Experiment

Most data analysis projects begin with a database.

This project began with a question.

Using ChatGPT, I explored:

• Driver behavior research

• Transportation studies

• Vehicle ownership demographics

• Vehicle design characteristics

• Highway safety data

• Psychological studies related to risk-taking and driving behavior

The AI reviewed available information and identified several recurring themes:

• Vehicle size and perceived safety

• Driver confidence

• Occupational usage

• Highway driving patterns

• Social and cultural influences

• Visibility advantages

• Risk compensation behavior

The result was not a definitive answer.

Instead, the result was a collection of evidence-based hypotheses.

Did AI Produce Valid Findings?

Yes and no.

The findings were useful because they highlighted recurring patterns found across multiple sources.

However, the findings were not statistically validated.

Without a structured dataset, AI cannot perform genuine statistical analysis.

It can identify patterns, summarize research, and generate theories, but it cannot prove causation.

This distinction is important.

Many people mistake AI-generated conclusions for analysis when they are actually informed summaries of existing information.

Below are the findings I was able to generate using AI, without access to a structured dataset or formal statistical analysis.

Pickup Trucks, Highway Behavior, and Why They Feel More Aggressive Today

This project began with a personal observation during my daily commute on Massachusetts highways. Over time, I noticed a recurring pattern that sparked a question and ultimately led me to explore whether AI could help uncover meaningful insights without access to a formal dataset. On both tighter highway stretches (where lanes feel compressed by barriers, exits, or shoulders) and wider interstates, I kept noticing a similar pattern:

Large pickup trucks—especially full-size models like the Ford F-150, Chevrolet Silverado 1500, and Ram 1500—often appear in the passing lane moving faster than surrounding traffic. In heavier congestion, they also seem more visible in any available “gap” space.

This isn’t limited to one road type; it shows up on both narrow-feeling highway sections and wide multi-lane systems. That raised a few questions I wanted to answer using data:

• Are pickups actually getting bigger?

• Are they getting faster?

• Are they more dangerous?

• And why does the behavior feel more noticeable now?

1. Are pickup trucks getting bigger?

Long-term size trend (all half-ton pickups averaged)

What the data shows

• Total growth since the 1980s: ~1.5–2 feet in length

• Width growth: only a few inches

• Height growth: about half a foot

Why they feel much larger today

The biggest change wasn’t gradual growth—it was structural design:

• Crew cab (4-door) became the default configuration

• Cab size expanded significantly

• Bed proportion became relatively smaller

• Front-end height and mass increased for safety and styling

So even with modest dimensional growth, the visual footprint increased more noticeably than the raw numbers suggest.

2. Are they getting faster?

Yes—especially in acceleration, not top speed.

Performance trend across eras

Supporting real-world data

Modern test datasets show:

• Average modern F-150: ~6.3 seconds 0–60 across trims

• Silverado 1500 average: ~6.8 seconds

• Comparable Ram 1500 performance in the same class (Consumer Reports comparison testing)

Key takeaway

Even though trucks gained weight, performance improved due to:

• turbocharged engines

• multi-speed transmissions (8–10 speed gearboxes)

• better torque delivery at low RPM

• early electrification in newer trims

3. Crash involvement (context, not blame)

National crash datasets from NHTSA show:

• “Light trucks” (including pickups) are involved in roughly 15–20% of all crashes

• They represent about 20–25% of fatal crashes

These are category-level statistics from:

• FARS (fatal crashes nationwide)

• CRSS (sampled police-reported crashes)

Importantly:

• This does NOT mean pickups are the most crash-prone vehicles

• It reflects a mix of exposure (miles driven) and crash severity

4. Why the behavior feels more noticeable on highways

A. Tight highway perception effect

On narrower-feeling roads:

• large vehicles visually dominate lanes

• following distance feels shorter

• speed differences appear more extreme

B. Traffic clustering in passing lanes

During commute hours:

• faster-moving traffic concentrates left

• pickups are now common commuter vehicles, not just work trucks

C. Vehicle capability change

Modern pickups:

• accelerate much faster than older generations

• maintain highway speeds with less effort

• merge and pass more decisively

D. Fleet size effect

Pickups are now a major share of new U.S. vehicle sales (multi-brand dominance), meaning:

• more of them are present in every traffic sample

• behavior becomes more visible statistically and visually

5. Are they more dangerous?

It depends on what “dangerous” means:

• Crash frequency: not unusually high per vehicle category

• Crash severity: higher than average passenger cars due to mass and geometry

So the correct interpretation is:

They are not uniquely crash-prone, but they are more overrepresented in severe crashes than smaller vehicles.

6. So what is actually happening?

Putting everything together:

• Trucks are slightly larger than past decades, but growth has slowed

• Trucks are significantly faster in acceleration than older generations

• Trucks are far more common as daily commuters

• Highway environments are more congested and visually compressed

• Safety statistics reflect severity more than frequency

So the observed behavior on highways is likely a combination of:

higher vehicle prevalence + faster acceleration + visual dominance in constrained road space

7. Simple forward projection (for fun)

Based on current trends (2010–2026 plateau in size + continued electrification in performance):

Projected averages

What drives the projection

• Size: likely plateaus due to road width and parking constraints

• Weight: increases slightly (especially with EV batteries), then stabilizes

• Speed: improves more noticeably due to electric torque and software control

Final takeaway

From a data perspective, what I’m observing on the road is consistent with long-term trends:

• Trucks are not dramatically bigger than they were 20 years ago—but they feel bigger because of design changes

• They are meaningfully faster in acceleration

• They are far more common as commuter vehicles

• And their visibility in traffic makes their behavior more noticeable than smaller cars

Data references

• NHTSA Fatality Analysis Reporting System (FARS): https://www.nhtsa.gov/research-data/fatality-analysis-reporting-system-fars

• NHTSA Crash Report Sampling System (CRSS): https://www.nhtsa.gov/crash-data-systems/crash-report-sampling-system-crss

• Ford F-150 specifications: https://www.ford.com/trucks/f150/

• Consumer Reports pickup comparisons (F-150 / Silverado / Ram): https://www.consumerreports.org/pickup-trucks/chevrolet-silverado-1500-vs-ford-f-150-vs-ram-1500-face-off-a9996190825/

• MotorTrend instrumented truck testing (acceleration comparisons): https://www.motortrend.com/reviews/2021-ford-f150-vs-chevrolet-silverado-ram-1500-pickup-truck-comparison-test-review/

• F-150 acceleration datasets (aggregated test data): https://www.0-60specs.com/ford/f-150-0-60-times

• Silverado 1500 acceleration datasets: https://www.0-60specs.com/chevrolet/silverado-1500-0-60-times

Not bad, but not valid.

The Generative AI was able to answer my questions, but without a clear dataset, the answers aren't valid. Even though I asked the AI to provide references, the findings cannot be considered fully validated.

What Would I Do Differently?

If I were performing this project as a formal data analysis exercise, I would start by building a dataset.

Potential data sources might include:

• Highway safety databases

• Insurance claims data

• Traffic citation records

• Vehicle registration data

• Driver demographic information

• Accident reports

Once collected, the data could be analyzed using Excel, SQL, Tableau, Power BI, or Python.

This would allow statistical testing instead of relying solely on AI-generated observations.

Could AI Generate the Dataset?

Technically, yes.

AI could help create a simulated dataset.

For example, ChatGPT could generate thousands of hypothetical vehicle records with attributes such as:

• Vehicle type

• Driver age

• Speed

• Traffic violations

• Accident frequency

The problem is that the resulting analysis would only be as good as the assumptions used to create the data.

A simulated dataset can be useful for practicing analysis techniques, but it cannot be used to make real-world conclusions.

The data would be artificial.

Does Using a Basic Program Like Excel Matter?

Absolutely.

Excel remains one of the most valuable tools in data analytics because it allows analysts to:

• Organize data

• Clean data

• Build calculations

• Create pivot tables

• Develop visualizations

• Validate assumptions

AI can assist with many of these tasks, but Excel provides transparency.

An analyst can trace every calculation, every formula, and every conclusion back to the source data.

That level of visibility is difficult to achieve when relying solely on AI-generated outputs.

My Conclusion

This experiment changed how I think about AI and analytics.

AI did not replace the role of a data analyst.

Instead, it acted as a research partner.

It helped gather information, identify patterns, challenge assumptions, and generate ideas for further investigation.

The experience reinforced an important lesson:

AI is excellent at helping analysts ask better questions.

But answering those questions still requires data, validation, and critical thinking.

In many ways, AI works best not as a replacement for analytics, but as an enhancement to the analytical process.