Peter Wallis, PhD

Formally of Melbourne, Australia, now living in the Peak District, half way between Manchester and Sheffield. I'm affiliated with Sheffield University and Emma is affiliated with the Centre for Policy Modelling in Manchester.



Email: initial.surname@dcs.shef.ac.uk
Office phone: 0114 222 1981
short bio and a CV
Peter's photo
I see myself as an AI person specializing in language and representation, but one who has moved on.   In the 1970's some really interesting observations were made about the nature of problem solving as search but that (IMHO) is a failed paradigm as it requires representation and that is problematic.   At first blush, the world appears to be full of things and words (representations) simply "stand in for" those things in some formalism.   It turns out that the way we group "stuff" into "things" could be far more arbitrary.   If our conceptual categories are not determined by the world, and categories are not as arbitrary as the Sapir-Whorf hypothesis would suggest, then how are they determined?

According to some real philosophers (not just me) it is pragmatics: Polar Bears are dangerous; some snow is good for building igloos; some snow is good for walking; some for sledding...   The semantics of oak tree DNA?   The argument, in too much detail, is made by Ruth Millikan and summarised by Hendriks-Jansen ["Catching Ourselves in the Act: Situated Activity, Interactive Emergence, Evolution, and Human Thought" MIT Press, 1996]. Others (among many) include Maggie Boden through to Steve Grand.   The conclusion?   Robotics is where the action is.   Rather than robotics done by engineers however, the real AI issues are being identified by those with the problems.   I have been working on robots for the military, robots as social actors, robots as physical entities, and robots as affective devices.   The point?   AI is really about us.   It is about the nature of being human in a world that is far more complex and evolved than our simple brains can cope with.   How do we do it?
Franco, the DSTO virtual assistant Embodied Conversational Agents (well, conversational systems and dialog really ...) Sarah with Harvey SERA - Social Engagement with Robots and Agents The Noo Noo A robot that grounds symbols in energy usage...more
Tethered in the back yard Ok, so the engineering is fun as well. Pneumatics has an energy to it that is a delight...more The SFPV under development UAVs - again fun, but sorting out the control!   The SFPV ended up with contra rotating blades in a shroud...more