The study, conducted by researchers at Queensland University of Technology, the University of the Sunshine Coast, Delft University of Technology and the Korea Advanced Institute of Science and Technology, surveyed 384 Australian adults aged 18 to 82. Participants evaluated fully connected and automated vehicles, or CAVs, before and after viewing an infographic describing their safety and justice dilemmas. In the study, a fully automated vehicle corresponds to Level 5 automation under the Society of Automotive Engineers’ classification: a vehicle capable of operating in all conditions without human intervention. Such a system would combine cameras, radar, lidar or other sensors with navigation software, artificial intelligence and communication links to vehicles, road infrastructure and digital networks. The technical ambition is continuous perception and decision-making, but the social question is who controls those decisions and who bears the consequences when they fail.

Awareness was remarkably high: 95.3 per cent of participants said they had previously heard of CAVs, while only 3.9 per cent reported no knowledge of them. Yet awareness was largely indirect. Media coverage was the dominant source, cited by 86.6 per cent of respondents, followed by conversations with friends, relatives or colleagues. Only 10.7 per cent had encountered the technology through public trials, and 7.4 per cent had experienced it through driving simulators. This imbalance matters because media narratives often compress complex engineering and governance questions into dramatic stories about crashes, futuristic convenience or technological breakthroughs. Most participants therefore approached CAVs with opinions formed before direct, sustained interaction with an automated system, rather than through practical experience of how a vehicle detects a pedestrian, negotiates uncertainty or communicates its intentions at a crossing.

The researchers measured attitudes on seven-point scales, asking about immediate feelings towards fully automated vehicles and the acceptability of associated safety and justice issues. After participants studied the infographic, positive perceptions declined slightly, but none of the changes was statistically significant. The mean score for immediate feelings fell from 4.39 to 4.31; the rating for the acceptability of safety issues dropped from 4.26 to 4.19; and the rating for the acceptability of justice issues declined from 4.32 to 4.26. Paired-samples t-tests indicated that these shifts could plausibly reflect chance variation rather than a meaningful effect of the information. In statistical terms, the intervention did not significantly change attitudes. In social terms, the result suggests that a single burst of information cannot easily dislodge deeper expectations about trust, responsibility and institutional competence.

Those expectations were especially visible in responses about justice. Participants were moderately positive about the possibility that safety and convenience could be distributed fairly, with an average score of 3.63, and about whether vulnerable groups’ needs would be considered, with a mean of 3.78. But confidence weakened when respondents considered equal access across income groups, which received a mean of 3.20, and whether benefits would be shared fairly, which scored 3.36. The concern is straightforward: advanced vehicles may be expensive to purchase, maintain and insure, while the public may still pay for upgraded roads, communications systems and traffic infrastructure. If the advantages of automation accrue mainly to wealthy urban users, CAVs could become a mobility technology that reinforces existing inequalities rather than reducing them.

Procedural and informational justice attracted even less confidence. Respondents gave relatively low ratings to community involvement in deployment decisions, opportunities to provide feedback and transparency in communication. Community participation scored 3.37, the opportunity to express feedback 3.23 and transparent communication 3.30. More than two-fifths of participants disagreed that communities were meaningfully involved or that communication was transparent. These findings point to a distinction often missed in technology policy: a system can be technically compliant while still lacking legitimacy. Procedural justice concerns who gets to shape deployment, whose experiences count during testing and how residents can challenge decisions. Informational justice concerns whether authorities and companies provide understandable information, respond to criticism and disclose limitations rather than presenting automation as an inevitable upgrade.

The open-ended responses showed that safety remained the most immediate concern. Among the 324 people who commented on safety, 29.8 per cent of coded responses focused on reliability, including software errors, sensor failures, hacking, loss of connectivity and poor performance in extreme or unpredictable conditions. Participants worried that automated systems might struggle to interpret the ambiguous cues humans use every day, such as eye contact, gestures and subtle movements at an intersection. Pedestrians, cyclists, older adults and people with disabilities were repeatedly identified as particularly vulnerable. A vehicle that accurately follows lane markings may still face difficult problems when a cyclist swerves, a child runs into the road, roadworks obscure signs or a pedestrian behaves in a way not represented in its training data. In such situations, the system must estimate uncertainty, predict other road users’ actions and choose a response within fractions of a second. That technical challenge becomes an ethical one when the distribution of risk is uneven.

Participants also questioned who would be responsible after a crash. In a conventional collision, responsibility may be assigned among drivers, vehicle owners, manufacturers, road authorities and other parties. With a CAV, agency is distributed across machine-learning models, sensor suppliers, software developers, fleet operators, telecommunications providers and infrastructure managers. A vehicle may be operating as designed while a faulty map, degraded sensor, cyberattack or unclear road marking contributes to harm. This diffusion of responsibility can make accountability difficult to trace. Respondents feared that pedestrians or cyclists could be blamed because “machines don’t make mistakes,” even though automated systems are designed, trained, tested and approved by human institutions. They also expressed distrust of private companies whose commercial incentives may not align with the public interest, particularly when software is proprietary and independent scrutiny is limited.

Infrastructure and geography added another layer of uncertainty in Australia. Fourteen-point-one per cent of coded broader concerns related to infrastructure, including road quality, communications networks, smart traffic signals and the readiness of rural and remote areas. Level 5 automation does not mean a vehicle is independent of its environment. Although a CAV’s onboard sensors can perceive the immediate surroundings, connected systems may depend on accurate digital maps, reliable positioning, machine-readable signs, network access and infrastructure capable of exchanging data. If those supporting layers are inconsistent, the vehicle may need to fall back on onboard perception or reduce its operating capability. Participants worried that costly upgrades would concentrate in major cities, leaving regional communities with fewer benefits and potentially greater exclusion. They also warned that road investment directed towards automated private vehicles could draw resources away from public transport, protected cycleways, safer crossings and accessible footpaths.

Environmental and economic concerns showed that participants viewed CAVs as part of a wider transport system rather than as isolated machines. Ten-point-three per cent of coded responses raised issues involving energy use, manufacturing, battery production, disposal, congestion and urban sprawl. Automation could reduce energy consumption per trip through smoother driving and better routing, but those gains might be overwhelmed if easier travel encourages longer commutes, more vehicle kilometres or a larger number of cars circulating without passengers. Another 6.8 per cent of responses focused on job displacement, especially among taxi, truck and other professional drivers. Participants called for retraining and support for workers whose livelihoods could be disrupted. Regulatory concerns accounted for 6.5 per cent, reflecting uncertainty over liability, oversight and the speed at which laws can adapt to software-based mobility. Privacy and cybersecurity, although less frequent at 2.1 per cent, carried significant implications: CAVs could continuously record location, travel patterns and interactions with infrastructure, creating a detailed data trail that companies, authorities or malicious actors might seek to access.

Despite this scepticism, the survey did not show blanket opposition. Support for continued deployment was comparatively stronger, with a mean score of 4.05, even as many respondents judged current arrangements weak on fairness and transparency. That combination amounts to conditional acceptance: people may be willing to entertain automated mobility if developers and governments can demonstrate reliable performance, protect vulnerable road users, make services affordable and establish clear lines of responsibility. The researchers caution that the sample was a voluntary online survey rather than a statistically representative portrait of all Australians. It may have favoured people with internet access, digital confidence or an interest in technology, and it included a higher proportion of participants reporting disability than the national census. The findings therefore cannot be treated as a precise national forecast. But they offer a clear warning for transport planners: public legitimacy will not be secured by technical capability or promotional messaging alone. The future of driverless vehicles will depend on whether institutions make risk visible, invite communities into decisions and ensure that automation serves mobility justice as well as engineering efficiency.