Bridging AI and Human Expertise: Designing for Trust and Adoption in Expert Systems by Stewart Smith

AI and Machine Learning are transforming expert systems, augmenting human decision-making in fields ranging from finance and healthcare to manufacturing and supply chain. But for AI to be truly effective, experts must trust and adopt these systems. This talk explores how UX practitioners can bridge the gap between AI’s computational power and human expertise.

We'll discuss key challenges, including designing for trust, working with the limits of explainability, and ensuring adoption through user-centered strategies. Attendees will gain practical insights into how to craft AI-driven experiences that experts rely on with confidence, ensuring these systems enhance rather than hinder decision-making.

1. May 9, 2025 STEW SMITH
Bridging Artificial
Intelligence and
Human Expertise

Designing for trust and adoption in expert systems.

2. May 9, 2025 STEW SMITH
Bridging Artificial Intelligence and Human Expertise

Designing for trust
and adoption in
expert systems.

3. AGENDA
Expert System Overview
Trust & Explainability
Design Strategies for Adoption
Questions & Discussion

4. May 9,
2025
STEW
SMITH
What is an
expert system?

5. May 9, 2025 STEW SMITH
An expert system is a type of computer program that
simulates the decision-making ability of a human expert.

It’s designed to solve complex problems by reasoning
through bodies of knowledge.

6. May 9, 2025 STEW SMITH
An expert system diagram Components
●Knowledge Base
●Inference Engine
●User Interface
●Explanation Facility*
●Knowledge Acquisition
Knowledge
Acquisition

7. May 9, 2025 STEW SMITH
Inference Engine Example Rule
IF engine does not start AND
battery is dead

THEN the problem is a dead
battery
The inference engine is
responsible for applying logic
to the knowledge base in
order to make decisions, draw
conclusions, or provide
recommendations - mimicking
expert reasoning.

8. May 9, 2025 STEW SMITH
Explanation Facility Purpose
Justify Conclusions (why, how)
Improve User Trust
Support Learning
Facilitate Debugging

The explanation facility in an
expert system serves a critical
role in building user trust,
transparency, and
understanding of the system’s
reasoning process.

9. May 9,
2025
STEW
SMITH
Trust &
Explainability

10. May 9, 2025 STEW SMITH
Trust is the cornerstone of system adoption.

Explainability is key to building trust.

11. May 9, 2025 STEW SMITH
All AI is Not Created Equal
Simple,
Decision Tree
Complex,
Large Language Model

12. May 9, 2025 STEW SMITH
Explainability, speed, and the “black boxˮ
The more complex the AI, the more difficult to explain.
INPUT to
OUTPUT
INPUT OUTPUT

13. May 9, 2025 STEW SMITH
The AI Trust Balance
Just Right
Over reliance

Under reliance
●Missed opportunities
●Increased cognitive load
●Reduced adoption and
engagement
●Loss of human judgment
●Inaccurate outcomes
●False sense of security

14. May 9,
2025
STEW
SMITH
Design Strategies

15. May 9, 2025 STEW SMITH
Common Trust Barriers Key Strategies
Uncertainty: Confidence scores and
uncertainty visualization.

Predictability: Ensuring consistent AI
behavior in recommendations.

Control: Allowing users to adjust or
override AI outputs.

Social Proof: Highlighting AI success
stories within an organization.
Lack of transparency in
decision-making (black box problem).

Perceived loss of control over key
decisions.

Fear of AI replacing human expertise
rather than augmenting it.

16. May 9, 2025 STEW SMITH
Adoption Challenge #1
Misalignment
between AI outputs
and existing
workflows.
Resistance to
change from
experts who have
deep domain
knowledge.
Adoption Challenge #2

17. May 9, 2025 STEW SMITH
Adoption Strategy for
MisalignmentResistance
Adoption Strategy for
• AI explanations must match
experts' mental models to improve
cognitive alignment.

• Present AI recommendations
within existing decision-making
workflows.

• Use domain-specific visualizations
to present data in a familiar way.
• Training & Onboarding: Helping
experts develop AI literacy.

• Providing clear affordances to
differentiate human and
AI-generated insights.

• Continuous feedback loops: Using
expert feedback to refine AI
recommendations.

18. May 9, 2025 STEW SMITH
AI isnʼt replacing human experts -
Itʼs empowering them to make
better decisions, faster
Final Thought

19. May 9, 2025 STEW SMITH
Trust is the foundation of AI adoption.

Explainability is a spectrum and must be balanced
with performance.

UX plays a critical role in bridging AI capabilities
and human expertise.
Key Takeaways Recap

20. Questions &
Discussion
May 9, 2025 STEW SMITH

21. THANK YOU!
May 9, 2025 STEW SMITH