The Life Scientific: Building Trust in Robots with Helen Hastie

Overview

In this episode of The Life Scientific, Helen Hastie, a professor of human robot interaction at the University of Edinburgh, talks through the challenges and opportunities of making robots that can work with people in everyday settings. She explains why trust in autonomous systems must be carefully calibrated and how robot appearance and function must align with the task at hand. The discussion covers a career spanning early dialogue systems to real-world social robots and national robotics initiatives, culminating in a vision of robotics integrated into daily life in trustworthy, human-friendly ways.

Key insights

• Uncanny valley and task-dependent appearance: robots should look human enough to be approachable, but not so close that disquiet distracts from their function.
• Trust calibration: designers must prevent both over trust and under trust, embedding guardrails and human oversight where appropriate.
• Real-world testing matters: field deployments, such as a robo barista and empathic tutors, reveal how appearance, dialogue style, and social cue understanding shape user attitudes and acceptance.
• Transparency and accountability: as robots undertake critical roles like triage or surgery assistance, systems should explain their reasoning and keep an audit trail.

Introduction

The podcast opens with an introduction to Helen Hastie, a leader in human robot interaction and head of Edinburgh's School of Informatics. Hastie outlines her mission to create robots that not only perform tasks competently but can also engage in natural conversations and explain their decisions to human colleagues. The conversation traces her career from the early days of spoken dialogue systems at AT&T Research Labs to contemporary social robots that can assist, teach, or triage, and to the National Robotarium, a collaborative UK facility aimed at translating research into real-world use.

From Dialogue Systems to Trustworthy Robots

The host and Hastie discuss common misperceptions about AI and robots. Hastie emphasizes that today’s robots are generally narrow in capability and that humanoid forms do not always confer advantages in real-world settings. She introduces the concept of the uncanny valley, noting that a robot’s facial features and expressions must be calibrated to the task. For home and service contexts such as factories or warehouses, functional design is more important than mimicking human form. The discussion also touches on the balance between reliability and user trust, noting that misalignment can lead to over or under trust in autonomous systems.

Early Work and Pivotal Projects