Enterprise AI Analysis
Trust in artificial intelligence: a survey experiment to assess trust in algorithmic decision-making
This study investigates public trust in AI-based Automated Decision-Making (ADM) through a survey experiment in Hungary (N=2100). Participants were presented with hypothetical decisions across four domains (medical, hiring, transportation, financial). The key finding is that human-assisted decisions are generally perceived as more trustworthy than ADM, except for financial investments. However, this effect is not uniform; factors like good AI understanding, low privacy concerns, and an open personality can mitigate negative impacts on trust. Sociodemographic and political variables had less influence. The study emphasizes the importance of transparency, accountability, and ethical design in fostering trust and adoption of AI systems.
Authors: Ferenc Orbán, Ádám Stefkovics | Publication Date: 31 March 2025
Executive Impact: Key Findings for Your Business
The study reveals critical insights into public perception of AI, offering strategic implications for enterprise AI adoption and trust-building.
2100
Participants Surveyed
4
Decision Domains Analyzed
-0.52
Avg. Trust Reduction by AI
3
Key Mitigating Factors
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
AI Ethics & Trust
Human-AI Interaction
User Perception & Acceptance
AI Ethics & Trust
The paper extensively covers the ethical dimensions of AI, specifically focusing on the concept of trust in Automated Decision-Making (ADM). It delves into why users might distrust AI, exploring factors like perceived bias, lack of transparency, and the 'black box' nature of algorithms. The core finding suggests that human-assisted decisions are generally more trusted than purely AI-driven ones, underscoring a fundamental human preference for human oversight in critical domains.
Relevance to Enterprise AI
For enterprises deploying AI, this highlights the necessity of transparent, explainable, and ethically designed AI systems. Building trust is paramount for adoption, especially in sensitive areas like HR, healthcare, or financial advice. Businesses must prioritize explainable AI (XAI) and human-in-the-loop models to counteract inherent distrust and ensure successful integration and user acceptance.
Human-AI Interaction
This section examines the interaction dynamics between human users and AI systems in decision-making contexts. It distinguishes between varying levels of automation, from AI assistance to full autonomy, and how these influence user perceptions. The study finds that 'low automation' scenarios, where AI supports human decisions, still face trust deficits compared to purely human processes, indicating a general wariness towards AI even in an assistive capacity.
Relevance to Enterprise AI
Enterprises should design AI interfaces that clearly communicate the role of AI (assistive vs. autonomous) and its limitations. User training and education about how AI functions can significantly improve perceived usefulness and trust. For critical enterprise applications, human oversight and clear decision pathways involving both AI insights and human judgment will be key to successful deployment and user adoption.
User Perception & Acceptance
The paper identifies several individual-level factors moderating trust in AI-based ADM. Notably, a good understanding of AI, low privacy concerns, and an open personality trait are found to mitigate the negative impact of AI assistance on trust. Conversely, traditional sociodemographic factors (age, gender, education, income, domicile) and political ideology showed less consistent influence on trust, suggesting that familiarity and personality are more dominant drivers.
Relevance to Enterprise AI
To enhance AI acceptance, enterprises should focus on user education and 'AI literacy' programs. Addressing privacy concerns through robust data protection measures and clear communication will be crucial. Furthermore, tailoring AI rollout strategies to acknowledge varying user personalities (e.g., more open individuals might be early adopters) could optimize adoption rates across the workforce and customer base.
-0.52
Average Trust Points Lower with AI (except finance)
Enterprise Process Flow
Good AI Understanding
→
Low Privacy Concerns
→
Open Personality Trait
→
Increased Trust in AI
| Domain | Human-Assisted Trust | AI-Assisted Trust | Key Trust Driver |
|---|---|---|---|
| Medical Diagnosis | High | Moderate (Negative Effect mitigated by AI understanding) |
|
| Hiring | Moderate | Low (Consistent Negative Effect) |
|
| Car Purchase | High | Moderate (Consistent Negative Effect) |
|
| Financial Investment | Low | Mixed (No Significant Difference, AI Understanding increases trust) |
|
Case Study: Financial Investment Trust
Unlike other domains, AI's involvement in financial investment decisions did not significantly decrease trust. In fact, for users with higher AI understanding, AI-assisted financial decisions were perceived as more trustworthy. This suggests a unique perception of AI in areas where data-driven optimization is highly valued and human limitations (e.g., emotional bias) are acknowledged.
Company: Fictitious Investment Firm
Industry: Financial Services
Challenge: Low human-assisted trust due to perceived biases and limitations in investment decisions.
Solution: Implemented an AI-driven investment advisor for fund allocation.
Result: No significant trust difference compared to human-only, and increased trust among users with high AI understanding, demonstrating AI's potential in data-heavy, objective domains.
Advanced ROI Calculator: Quantify Your AI Advantage
Input your operational data to see estimated annual savings and reclaimed human hours by integrating intelligent automation.
$5,200,000
Estimated Annual Savings
104,000
Human Hours Reclaimed
*Estimates based on industry benchmarks and average AI efficiency gains.
Your AI Implementation Roadmap
A structured approach to integrating AI, ensuring measurable impact and seamless adoption within your enterprise.
AI-assisted decisions are generally perceived as less trustworthy than human-assisted decisions, with the exception of financial investments.
Higher AI understanding, lower privacy concerns, and an open personality can significantly mitigate negative trust impacts of AI assistance.
Sociodemographic (age, gender, education, income, domicile) and political factors showed minimal influence on trust in AI-based ADM.
Trust levels varied by domain, with medical and transportation domains showing higher trust than hiring and financial domains.
The study highlights the crucial role of transparency, accountability, and ethical design in building public trust and facilitating AI adoption.
Ready to Transform Your Enterprise with AI?
Unlock the full potential of artificial intelligence. Our experts are ready to design a tailored strategy for your business.
Ready to Get Started?
Book Your Free Consultation.
Let's Discuss Your AI Strategy!
Lets Discuss Your Needs
Select a Date
Sun
Mon
Tue
Wed
Thu
Fri
Sat