Wednesday, Aug 5: 2:00 PM - 3:50 PM
1056
Invited Paper Session
Thomas M. Menino Convention & Exhibition Center
Room: CC-258C
Applied
Yes
Main Sponsor
Business and Economic Statistics Section
Co Sponsors
Section on Statistical Learning and Data Science
Presentations
The growing field of interpretable machine learning promises to open the black box of modern predictive algorithms, offering human-understandable explanations for complex models. Yet the meaning, validity, and social consequences of these "explanations" remain deeply uncertain. This talk surveys major approaches to interpretability—from local feature attributions and counterfactual reasoning to surrogate modeling and causal abstraction—and asks what, if anything, they tell us about how models actually behave. We will examine how explanations interact with bias and uncertainty, where uncertainty arises, and how adversarial examples and prompts exploit these weaknesses.
Beyond their statistical properties, explanations that are provided to the subjects of automated decisions also affect subject behavior, inducing distribution shift with ramifications for both model performance and explanation validity.
Keywords
Machine Learning
Explainable AI
Interpretable Models
Distribution Shift
Can AI alone drive scientific discovery, or does the future belong to human-AI collaboration?
This talk argues that while AI systems show remarkable capabilities in data analysis and
pattern recognition, carefully designed human-AI co-creation likely outperforms either
working independently in expert tasks.
We present evidence that AI-only approaches face critical limitations: lack of contextual
understanding, difficulty with problem formulation, and inability to ensure population
alignment. However, when humans and AI collaborate effectively—with AI handling computational
heavy lifting while humans provide domain expertise, ethical oversight, and interpretive
judgment—the combination achieves what neither can accomplish alone.
Keywords
AI
data driven
pattern
context
AI aims to get computers to do what humans can do:
To reason, plan, exhibit goal oriented actions,
all enabled by processing images, speech and language
for pattern recognition.
Computer 'Learning' (CL) algorithms do this?
But other disciplines occupied these research/knowledge
territories long before and AI-CL lacks their expertise.
We will:
deconstruct some of the hype (e.g.computer not machine);
demonstrate the vast gulf between human capabilities and computers;
exemplify the articulation of the real expertises;
and argue that anthropomorphic language is dangerous and we must stop using it.
What is left are some promising tools;
but which cannot function reliably without human guidance.
Keywords
skeptical AI
anthropomorphic
intelligence gulf