The Gradient: Perspectives on AI

Vera Liao: AI Explainability and Transparency

Dec 7, 2023

01:37:03

Snipd AI

Vera Liao, Principal Researcher at Microsoft Research, discusses AI explainability and transparency. Topics include UX design and toolkits for AI explainability, interfaces for explanation and model capabilities, online social communities, the difference between explainability and interpretability, the importance of interactivity in explanations, understanding the social context of AI systems, evaluating AI explainability criteria, challenges in AI transparency, and the need for a collaborative community.

AI Summary

Highlights

AI Chapters

Episode notes

Podcast summary created with Snipd AI

Quick takeaways

The responsible design of AI systems requires transparency and explainability to gain insights into specific usage contexts.

Vera Liao focuses on human interaction and understanding to bridge the gap between AI capabilities and user needs.

Exploring the social-technical system of AI, explanations can support human understanding and decision-making by considering social context and interaction traces from other users.

Deep dives

The Importance of Transparent and Explainable AI Systems

The responsible design of AI systems requires not only high-performing algorithms, but also transparency and explainability. This allows users to gain insights into the models based on their specific usage contexts. Vera Liao, a principal scientist at Microsoft Research, has been studying explainability and transparency in AI systems. Her research aims to develop explainability solutions that meet the diverse needs of different users. By understanding the limitations of existing explainability methods, she explores ways to integrate human interaction and understanding into AI systems. Additionally, she discusses the challenges and potential of transparency in the era of large language models.

Scoring Accuracy Motivation on the Spectrum

03:07

How to Slow People Down in Decision-Making

04:35

Using a 4W Framework to Understand Interactions and Decisions

00:48

Introduction

2min

HCI Research on AI Technology and User Interaction

15min

Exploring Filter Bubbles and Echo Chambers and Designing for Receptiveness to Different Opinions

5min

Influencing Open-Mindedness in Tech and Interface Design

2min

Explainability and Transparency in AI

16min

The Importance of Interactivity in Explanations

2min

Social Transparency: Understanding the Broader Context

8min

Evaluating AI Explainability: Perspectives and Criteria

21min

The Importance of AI Transparency and Understanding

6min

10.

The Challenge of Transparency in AI Models

16min

11.

Supporting the Ecosystem and Building a Collaborative Community

4min

In episode 101 of The Gradient Podcast, Daniel Bashir speaks to Vera Liao.

Vera is a Principal Researcher at Microsoft Research (MSR) Montréal where she is part of the FATE (Fairness, Accountability, Transparency, and Ethics) group. She is trained in human-computer interaction research and works on human-AI interaction, currently focusing on explainable AI and responsible AI. She aims to bridge emerging AI technologies and human-centered design practices, and use both qualitative and quantitative methods to generate recommendations for technology design. Before joining MSR, Vera worked at IBM TJ Watson Research Center, and her work contributed to IBM products such as AI Explainability 360, Uncertainty Quantification 360, and Watson Assistant.

Have suggestions for future podcast guests (or other feedback)? Let us know here or reach us at editor@thegradient.pub

Subscribe to The Gradient Podcast: Apple Podcasts | Spotify | Pocket Casts | RSSFollow The Gradient on Twitter

Outline:

* (00:00) Intro

* (01:41) Vera’s background

* (07:15) The sociotechnical gap

* (09:00) UX design and toolkits for AI explainability

* (10:50) HCI, explainability, etc. as “separate concerns” from core AI reseaarch

* (15:07) Interfaces for explanation and model capabilities

* (16:55) Vera’s earlier studies of online social communities

* (22:10) Technologies and user behavior

* (23:45) Explainability vs. interpretability, transparency