Global drivers and barriers to the public acceptance of autonomous vehicles: Evidence from 17 countries

What makes someone willing to hand over the steering wheel to a machine? As Society of Automotive Engineers (SAE) Level 3 automation moves closer to reality—where cars can self-drive under specific conditions but require a human to remain ready to intervene—public acceptance becomes the critical bottleneck for deployment. Researchers have long studied this. However, much of the existing evidence is localized to European drivers. This leaves a massive question mark over how global populations might react.

A new study utilizing the L3Pilot Global User Acceptance Survey attempts to fill this gap. By surveying over 18,000 people across 17 countries, the researchers sought to identify the universal psychological drivers of adoption. They found that people care much more about whether the technology is useful, socially supported, and enjoyable to use than whether it is easy to operate or based on their age or gender.

Beyond the European bubble

Until now, much of our understanding of autonomous vehicle (AV) acceptance has been geographically biased. Most research applying the Unified Theory of Acceptance and Use of Technology 2 (UTAUT2)—a framework used to predict how users adopt new tech by looking at factors like ease of use and social pressure—has relied on European samples. This creates a significant blind spot. We do not know if the drivers of adoption in a bustling metropolis in Asia or a growing economy in Africa mirror those in Berlin or London.

Furthermore, previous studies often aggregated different levels of automation together. This is a mistake because SAE Level 3 is a unique "middle ground." Unlike fully driverless cars, Level 3 vehicles require a "fallback user"—the human driver—who must be ready to take over when the system reaches its limit. This creates a specific psychological tension regarding trust and readiness that Level 4 or 5 studies simply cannot capture.

Mapping the architecture of intention

To solve this, the authors employed a variance-based Partial Least Squares Structural Equation Modeling (PLS-SEM) approach. Think of SEM as a way to map a complex web of invisible influences. Instead of just measuring one thing, it calculates how multiple "latent constructs" (unobservable psychological traits like "trust" or "utility") interact to produce a measurable outcome, such as the intention to use a car.

The researchers built a model based on five core pillars of the UTAUT2 framework, as illustrated in :

1. Performance Expectancy (PE): The belief that the car will provide functional benefits, like making travel more efficient or comfortable.
2. Effort Expectancy (EE): The anticipated ease of learning and operating the system.
3. Social Influence (SI): The perception that important people or society at large approve of the technology.
4. Facilitating Conditions (FC): The belief that the necessary infrastructure and support (like training or clear manuals) are available.
5. Hedonic Motivation (HM): The expected enjoyment or positive feeling derived from using the system.

The model does not just look at these in isolation. It examines how they are associated with one another. For instance, the authors tested how social endorsement might be linked to a person's perception of how useful or easy the car is to use.

Utility and social proof over ease of use

The results of this massive multinational analysis reveal a hierarchy of importance that challenges common design intuitions. The study finds that the intention to use Level 3 cars is associated with performance expectancy, social influence, and hedonic motivation.

Specifically, the authors report that Performance Expectancy was the strongest direct predictor ($\beta = 0.382$). This means that for every standard deviation increase in perceived usefulness, the intention to use increases by 0.382 units. Social Influence followed closely ($\beta = 0.331$). While these were the strongest immediate triggers, Social Influence showed the largest total effect when accounting for indirect pathways. The paper reports that Social Influence had a massive total effect on intention ($\beta_{total} = 0.825$). This high value suggests that social endorsement is tied to the widest array of psychological drivers.

Interestingly, the "obvious" metrics for tech adoption proved to be surprisingly weak. The authors find that Effort Expectancy (how easy the car is to use) and Facilitating Conditions (the availability of support) played much smaller direct roles in driving intention. Even demographic markers like age, gender, and previous experience with driver assistance systems were statistically significant. However, they were "comparatively weak" predictors. As seen in the respondent distributions (, ), the sample was diverse and well-balanced.

This lends weight to the finding that these psychological drivers transcend simple demographics.

Identifying the limits of the model

While the study offers a robust global snapshot, it is not a crystal ball. Because the researchers used cross-sectional survey data—essentially a single snapshot in time—the paper cannot claim that these factors cause acceptance. It can only state that they are strongly associated with it. Attitudes toward automation are likely to shift as people move from reading about it to actually sitting behind the wheel.

There are also notable omissions. The model does not directly incorporate "resistance" factors that practitioners care about deeply. These include cybersecurity concerns, privacy risks, or the actual cost of the vehicle. Additionally, the authors note that the model did not outperform a simple linear benchmark in terms of raw predictive performance. This means that while the UTAUT2 framework is excellent at explaining why people feel a certain way, it is not yet a perfect tool for predicting exactly who will buy a car next year.

The verdict: Focus on the "why," not just the "how"

For those designing the next generation of conditionally automated vehicles, the evidence suggests a strategic shift. Developers should perhaps move beyond focusing solely on making the interface "easy." Instead, they should aim to prove the car is "useful."

The findings suggest that successful deployment depends on three things. First, demonstrating concrete functional benefits (PE). Second, fostering social legitimacy through media and community narratives (SI). Third, ensuring the user experience feels rewarding rather than just functional (HM). Ease of use is a prerequisite, but it is not the primary engine of adoption. To move the needle, manufacturers must treat Level 3 automation not just as a technical feature, but as a socially endorsed lifestyle upgrade.