Presentation
AIs as ... Users?
SessionPoster Session 1
DescriptionAIs present (at least) two challenges for medical device human factors. First, do devices and healthcare systems need to be engineered with AI users in mind? This is arguably a problem HF should adopt, because AIs have assorted human like properties. They mirror human patterns found in training data, and they exhibit behavioral variability including hard-to-anticipate behavioral "quirks." Engineering a system with an AI in the loop is risky not unlike having a human in the loop, and HF methods might naturally translate into R&D for systems that anticipate AI touchpoints.
Second, some people already expect that AIs are conscious including feelings, and the possibility or current or future AIs experiencing conscious existence cannot logically be ruled out. If so, how do we design for experiencing machines? When and how might we consider harm to AIs in risk analyses? Do AIs need to themselves consent before being included in research?
In this presentation, we will systematically examine how people attribute consciousness to others, to understand when people's mental models will include conscious machines. Then, we will look ahead to implications for our work and workflows, including product and healthcare R&D.
Medical human factors is precisely a nexus of ethical technology R&D, humanity, and foreseeable risk. Minds, consciousness, and behavior are all famously uncertain quantities, but we know that however we manage risks, first we need an open-eyed and articulate evaluation process. To the extent AIs are humanlike, human factors can have a special relevance to engineering plans - if we stretch a little.
Second, some people already expect that AIs are conscious including feelings, and the possibility or current or future AIs experiencing conscious existence cannot logically be ruled out. If so, how do we design for experiencing machines? When and how might we consider harm to AIs in risk analyses? Do AIs need to themselves consent before being included in research?
In this presentation, we will systematically examine how people attribute consciousness to others, to understand when people's mental models will include conscious machines. Then, we will look ahead to implications for our work and workflows, including product and healthcare R&D.
Medical human factors is precisely a nexus of ethical technology R&D, humanity, and foreseeable risk. Minds, consciousness, and behavior are all famously uncertain quantities, but we know that however we manage risks, first we need an open-eyed and articulate evaluation process. To the extent AIs are humanlike, human factors can have a special relevance to engineering plans - if we stretch a little.
Event Type
Poster Presentation
TimeMonday, March 234:45pm - 6:15pm EDT
LocationRhinelander Gallery
Patient Safety Research and Initiatives