Closing Keynote
This talk opens with a formal model of conversation as an offering to designers of conversational user interfaces. Based on that model, it proposes minimal requirements for interfaces that are collaborative, ethical, and humane. These apply to human-machine, human-human, and machine-machine interactions. Lastly the talk uses a seminal installation of physical computing from 1968 to ask the question, what do we want from our conversations with machines?
The technology of chatbots is not old. The first chatbot, Weizenbaum’s ELIZA, is only 50 years old. But the technology of human conversation — the countless daily interactions that enable learning, respect, trust, collaboration, and society — is something like 2,000 times older. Human conversation is far more mature than anything digital, and evolution has tuned it to our biology.
As we design interactive systems, should we forefront the technology of chatbots or the technology of human conversation?
{And then there is AI, a key technology for chatbots. Every day AI is inside more of the apps we use and more of the systems that control our physical environment. Soon AI will be “in the air we breathe”, delivered by speech interfaces everywhere we go. But just like chatbots, AI takes its lead from digital technology — how computers work and how they represent the problems we want them to solve. (Neural nets are a sliver of nervous system functioning, so we can’t claim they are terribly human.) AI is not rooted in humanism or social good. That’s why platform companies like Google and non-profits like The Partnership on AI are playing catch-up and composing ethical principles — because AI is not fundamentally ethical or fundamentally humane.}
Adding a sweet, ethical icing to the bitter consequences of digitalizing human conversation feels perilous. Rather than be led by the technology of chatbots or AI, can we forge a path toward conversational systems that are fundamentally natural, humane, social, and ethical?
What are more of the questions should we be asking? And who should be in our conversation?DownloadPPresentation Slides [50 MB]
DownloadPAudio Recording (apologies for poor audio)
ViewCOLLOQUY 2018 Project Video on Vimeo
DownloadStudent Website for COLLOQUY 2018 Project
DownloadConversation and Interfaces
DownloadBlog Posts on Conversation & Design
DownloadLinks about Designing for Conversation
ViewPangaro's RSD5 Video on Vimeo
Special thanks to Caio Calado, Hugh Dubberly, TJ McLeish, and Karen Berntsen.
Paul Pangaro has been designing conversational interfaces for 40 years. At MIT he received a BS degree in Humanities/Computer Science and then was hired by Nicholas Negroponte to become a member of the research staff of the MIT Architecture Machine Group, predecessor of the MIT Media Lab. There Pangaro met Gordon Pask, who had for 20 years been making interactive systems for entertainment, research, and training, from which he developed conversation theory, a comprehensive and rigorous framework for understanding and fostering conversation, in machine-machine, human-machine, or human-human interactions. Pangaro earned a PhD in Cybernetics with Pask at Brunel University (UK), pursuing a career as entrepreneur, teacher, researcher, and consultant. He has worked with and within software startups in New York, Boston, and Silicon Valley in product and technology roles. As a consultant, Pangaro has been engaged by Du Pont, Nokia, Samsung, Instituto Itaú Cultural (São Paulo), Ogilvy & Mather, 8 Inc, Intellectual Ventures, and PoetryFoundation.org. His published papers explicate “designing for conversation" from his research and his implementations of software products and organizational processes. His most recent project is the full-scale replication of Pask’s Colloquy of Mobiles at the College for Creative Studies in Detroit, where he was Chair and Associate Professor of the MFA Interaction Design program from April 2015 through December 2018. In January 2019 Pangaro will begin a new role at Carnegie Mellon University as Professor of the Practice in the Human-Computer Interaction Institute.