Enterprise AI Analysis
Not All Trust is the Same: Effects of Decision Workflow and Explanations in Human-AI Decision Making
This study investigates the complex relationship between decision workflow (1-step vs. 2-step), AI explanations, and user characteristics on trust and reliance in AI-assisted decision making. We conducted an online study (N=300) in a fictional university student support scenario. Key findings indicate that self-reported trust and behavioral reliance are distinct constructs, only weakly correlated. Contrary to expectations, a 2-step setup increased overreliance. The effectiveness of explanations was context-dependent, boosting trust in a 2-step setup but lowering it in a 1-step setup. Domain knowledge also significantly influenced reported trust. These results highlight that trust in AI is not monolithic and its calibration requires careful consideration of workflow design and user context.
Executive Impact Snapshot
Key findings highlight critical considerations for designing trustworthy and effective human-AI decision support systems across the enterprise.
0
Participants Studied
0%
AI Accuracy
0%
User Agreement Rate
0 min
Avg. Study Time
Deep Analysis & Enterprise Applications
Select a topic to dive deeper, then explore the specific findings from the research, rebuilt as interactive, enterprise-focused modules.
Trust Measures
Decision Workflow Effects
Explanations & Interactions
Domain Knowledge & Experience
Weak Correlation
Reported Trust vs. Behavioral Reliance
This study found only a weak correlation (r = .25) between self-reported trust (HCT questionnaire) and behavioral reliance (agreement rate or switch rate), indicating they are distinct constructs.
2-Step Decision Workflow Process
Decision Task Presented
→
User Submits First Decision
→
AI Suggestion & Explanation Displayed
→
User Submits Final Decision
Increased
Overreliance in 2-Step Setup
Contrary to previous suggestions, the 2-step decision setup significantly increased overreliance (accepting incorrect AI advice) compared to the 1-step setup (β = 0.46, p = .001), despite taking more time.
| Characteristic | 1-Step Workflow | 2-Step Workflow |
|---|---|---|
| AI Suggestion Timing | Immediate with Task | After User's Initial Decision |
| Reported Trust (with explanations) | Slightly Lower (M=3.09) | Higher (M=3.34) |
| Reported Trust (no explanations) | Slightly Higher (M=3.25) | Slightly Lower (M=3.20) |
| Overreliance Outcome | Lower | Significantly Higher |
| Avg. Time per Study | 13.8 minutes | 16.4 minutes |
| Switch Rate Measurable | No | Yes |
Context-Dependent
Explanation Effectiveness
Explanations had a crossover interaction effect: they raised reported trust in a 2-step setup but lowered it in a 1-step setup. This suggests explanation impact is not universally positive and depends on the decision workflow.
University Student Support Task & User Characteristics
Scenario: Participants acted as student support officers, deciding the academic trajectory (graduate/dropout) of second-year university students based on provided data and AI predictions.
Key Findings:
- AI accuracy was 73%, matching human accuracy at 73%.
- Users with low domain knowledge reported significantly lower trust in the AI, though their behavioral reliance did not differ.
- Domain knowledge positively predicted reported trust (β = 0.24, p = .001).
- Interaction effects observed between workflow and domain knowledge on reported trust.
Calculate Your Potential AI ROI
Estimate the efficiency gains and cost savings your enterprise could achieve by implementing intelligent automation solutions.
Estimated Annual Savings
$0
Hours Reclaimed Annually
0
Your AI Implementation Roadmap
A structured approach to integrating AI into your enterprise, ensuring calibrated trust and maximum operational efficiency.
Discovery & Strategy
Identify high-impact areas, define objectives, and assess current infrastructure. Focus on understanding existing decision workflows and potential AI integration points.
Pilot Program & Workflow Design
Develop a focused pilot AI-assisted system. Experiment with different decision workflows (1-step vs. 2-step) and explanation types to understand user trust and reliance in your specific context.
User Training & Calibration
Implement targeted training programs for users, emphasizing how to interpret AI advice, understand AI limitations, and calibrate their trust based on domain knowledge and system explanations.
Iterative Deployment & Monitoring
Roll out AI solutions incrementally, continuously monitoring performance, user feedback, and trust metrics. Adapt workflows and explanation strategies based on real-world usage and evolving needs.
Scaling & Optimization
Expand successful AI implementations across the enterprise, optimizing for performance, user experience, and sustained trust. Explore adaptive workflows that tailor AI interaction to individual user profiles and task contexts.
Ready to Optimize Your Enterprise with AI?
Leverage our insights to build AI solutions that your team can truly trust. Book a consultation to discuss a tailored strategy for your organization.
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