Understanding Operator Attitudes Toward AI-Supported Decision Making in Maritime Operations
Doreen Jirak, Armeen Saroukanoff, Dirk van Rooy
cs.HC · cs.AI
Quick verdict
Survey of 66 seafarers finds positive AI attitudes and stable trust across scenarios; valuable expert data, but null age/experience results are underpowered.
Summary
This paper reports a survey study of maritime professionals' attitudes toward an AI-supported decision assistant in collision-avoidance scenarios. A Qualtrics survey disseminated via the Antwerp Maritime Academy and its alumni network was accessed by 166 people, with 66 complete responses retained, roughly half from Captains or Officers of the Watch. Participants assessed two simulated (Wizard-of-Oz) collision-avoidance scenarios — one unambiguous, one ambiguous with dense traffic — shown as ECDIS/AIS images, first unaided, then with an XAI overlay displaying speed and rudder feature importance. They completed the ATAS technology-anxiety scale (α=0.86), the Jian Trust-in-Automation scale (α=0.910/0.936), and a modified Hoffman XAI scale (α=0.633/0.609; factor analysis revealed three latent factors), plus open-ended questions analyzed via sentiment and thematic analysis. Results indicate low technology anxiety, stable trust across scenarios (paired t-test p=0.936), no significant age or sea-experience differences (Kruskal-Wallis), and scenario-sensitive explanation ratings; open responses highlighted confidence support and situation awareness as benefits, and over-reliance, skill degradation, and input reliability as concerns. Limitations include a 40% completion rate, a small and unevenly distributed sample, static image stimuli, and the absence of effect sizes, confidence intervals, and multiple-comparison corrections.
Review complete
Assessment at a glance
5
risk flags identified
● high 0
● medium 3
● low 2
5 strengths · 5 weaknesses · 5 author questions
AI text signal = Binoculars zero-shot detector score over 20 sampled passages (Fast-DetectGPT co-reported) — absence does not prove human authorship, and a flag does not imply misconduct.
Levels reflect verification complexity — not a quality score. Evolve never rates your paper.
Key claims under review
- Maritime professionals show generally positive technology disposition (ATAS median 2) with no significant differences across age or sea-experience groups.
- Trust in the simulated assistant remained stable between an unambiguous and an ambiguous collision-avoidance scenario (paired t-test p=0.936).
- Explanation ratings are multidimensional and scenario-sensitive (Hoffman scale α=0.609-0.633, three latent factors), and participants were unlikely to change decisions based on the agent's output (item 8 median 2).
- Open responses value decision support, situation awareness, and confidence building while flagging over-reliance, skill degradation, and input reliability (Tables 2-3).
- Maritime AI should target calibrated reliance through transparent, reliable, operationally meaningful design with domain experts in the loop (§4-5).
Strengths
Hard-to-reach expert sample
66 complete responses with roughly half Captains/OOWs, recruited via the Antwerp Maritime Academy and alumni (Table 1); instrument pilot-tested with 10 professional seafarers before dissemination (§2).
Domain-grounded scenario design
ECDIS/AIS images from maritime company MAHI, ground truth and COLREG interpretations authored by Cpt. Saroukanoff, and a realistic bridge framing (Master arriving to an unclear close-quarters situation, §2.1).
Honest psychometric reporting
Informative mixed-methods synthesis
Open materials and candid limitations
Weaknesses
Small, skewed sample with heavy attrition
Null claims without power analysis or effect sizes
No multiple-comparison correction
Weak-scale composite still tested
Underspecified qualitative pipeline
Full Review Report
$1.99 · one-timeUnlock the complete preliminary analysis Evolve produced for this paper.
Report preview
Scope & Claims
The paper reports a questionnaire study (HLDM@ECMLPKDD 2026 pre-print) of how maritime professionals perceive an AI-supported 'maritime assistant' in two simulated collision-avoidance scenarios. It addresses three RQs (§2): technology disposition (RQ1, ATAS), the effect of explainability on satisfaction, trust, and adoption (RQ2, TiA plus a modified Hoffman scale), and open-ended expectations and concerns (RQ3). Central claims: generally positive technology attitudes with no age or sea-experience gradients (§3.1); trust stable across an unambiguous and an ambiguous scenario (paired t-test p=0.936); explanation ratings multidimensional and scenario-sensitive (α=0.633/0.609, three latent factors); and a qualitative picture in which confidence support and situation awareness are welcomed while over-reliance, skill degradation, and input reliability dominate concerns (Tables 2-3). The prescriptive claim — design for calibrated reliance rather than maximal trust (§4-5) — follows from the data but is also a standing XAI position the survey only partly tests.
The remaining sections cover Methodology Assessment, Statistical Evidence, Results & Reproducibility, Limitations & Risks, and Verdict & Recommendations.
- All 5 strengths & 5 weaknesses, analyzed in detail
- 5 risk flags with severity, section & reviewer questions
- Methods & statistical evidence assessment
- 5 questions a peer reviewer would ask
- Full six-dimension evaluation report & recommendations
One-time purchase · this paper only
One-time purchase, this paper only. AI-generated preliminary analysis — not peer review.
Frequently asked questions
Did the study find that older seafarers trust maritime AI less?
No. Kruskal-Wallis tests found no statistically significant differences in technology anxiety (ATAS) or trust in automation by age or sea-experience group; the closest result was trust by age in scenario 2 (H=8.969, p=0.062). With only 66 respondents in five uneven groups, the authors themselves state the data do not allow statistically robust conclusions about age or experience effects.
Was a real AI system used in the survey?
No. It was a Wizard-of-Oz simulation: static ECDIS and AIS images of two collision-avoidance scenarios, first shown unaided, then with authored 'XAI' overlays displaying speed and rudder importance scores. The authors note no deceptive agent behavior was introduced, so all trust judgments concern a simulated assistant.
How many maritime professionals participated and who were they?
166 people accessed the survey and 66 completed it (about 40%), collected from December 2025 to mid-April 2026 via the Antwerp Maritime Academy and its alumni network. Roughly half were Captains or Officers of the Watch; about half reported more than 72 months of sea service.
What were seafarers' main concerns about AI collision-avoidance assistants?
Their open-ended answers flagged over-reliance and skill degradation (especially among junior officers), unreliable inputs such as GPS spoofing, jamming, or sensor failure, information overload and alarm burden, and doubt about handling complex sequential traffic. They valued decision support, situation awareness organization, and confidence building for younger officers.
About this review
This is an AI-generated preliminary review by Evolve.Pub, produced from the openly available text of the paper. It is not peer review and does not represent the authors' views. The canonical record of the paper is at the publisher link above.
- Figure panels (Figs. 1-11, including the Fig. 9 scale-item wording) are graphical; only their textual descriptions in the body were available, so figure-level values could not be independently checked against the reported statistics.
Go deeper on this paper
Wrote a paper? Stress-test it before reviewers do.
Upload your manuscript for the same structured scrutiny — reasoning, statistics, evidence, figures, and reproducibility.