Natural intelligence, as observed in humans and animals, is the result of multiple systems and subsystems, implementing a complex pattern of information flow and controlled interaction. How can complex interactions between a vast number of largely self-organized functions produce a thinking mind? This question is first and last an architectural one: how the system operates as a whole. Without a deep understanding of architecture we will never understand intuition, attention, insight, or understand understanding itself. My focus is on how the architecture of mind can be implemented in an artificial substrate. By building working – running – models of mind we may be able to kill two flies in one swat: understand the mind while and build an artificial general intelligence that can be used for a myriad of tasks, from designing new clothing to cleaning up our attic to helping solve global warming.

I try to build larger, more integrated and complete systems than achieved to date, because I find it unlikely that mind appears from a simple principle — rather, I think that the mind is the result of a huge amount of interacting components, hooked up in a very complex way according to largely unknown principles. My approach follows two main traditions in systems thinking. On the one hand is a rather familiar modular decomposition from cognitive science and software development. Modularization (object-orientation being one expression) is the most highly advanced method at present to construct complex systems - constructionist A.I. Unfortunately this method has severe limitations. As the proponents of the holistic systems approach have pointed out (e.g. Varela, Maturana, Simon) many complex systems have the elusive property that local interactions between their parts are not sufficient to explain, understand or predict the operation of the whole system of which they are part. Software methodologies employing traditional modular decomposition will not be sufficient to allow us to construct such systems in the lab.

If we are ever to see generally intelligent artificial systems we must look towards methodologies that more directly allow us to model and study complex phenomena, calling for an investigation of the principles of self-organization and meta-control. In short, we must employ methods that allow the system to develop on its own, through self-constructive principles. This is constructivist A.I. This topic is the subject of my 2009 AAAI Fall Symposium on Biologically-Inspired Cognitive Architectures keynote speech, as well as the subject of the HUMANOBS Workshop From Constructionist to Constructivist AI held in the fall of 2011, and one of the main topics of our upcoming Summer School on Constructivist AI and Artificial General Intelligence.

Selected Projects

Constructivist A.I.

A.I. research on applying principles of self-organization in the development, implementation or construction of A.I. systems is called constructivist A.I. As it is becoming clear that the manual construction process employed in most of software development wil not be sufficient to construct the kinds of complex architectures that we require for general intelligence, our focus must shift towards using techniques that allow systems to acquire their own knowledge and grow on their own. Without such principles in hand it is unlikely that we will we see systems with architecture-wide integration of learning, attention, analogy making and system growth. Our recently-awarded HUMANOBS project grant from the EU will enable us to take notable steps in this direction.

Constructivist Papers
From Constructionist to Constructivist A.I.
Self-Programming: Operationalizing Autonomy
Achieving Artificial General Intelligence Through Peewee Granularity

Constructionist A.I.

Constructionist A.I. (not to be confused with constructivist A.I. - see above) is a moniker given to the bulk of A.I. research being performed around the world, where traditional software development methods form the basis of the work. In this tradition we have developed the Constructionist Design Methodology (CDM), which eases the creation of modular, complex machines that incorporate some aspects of a full perception-action loop. We have used it on the HONDA Asimo humanoid robot and Mirage autonomus virtual agent [ watch movie]. Mirage inhabits an augmented reality; this complex system of integrated heterogeneous components was designed and implemented in as little as 2 mind-months using the CDM. We think it's directly due to the application of CDM in the project [published in A.I. Magazine, winter 2004]. We have also used CDM for a live performance of the Robot Opera in Reykjavik, 2006 [ watch movie] and 2007. One of our main and ongoing projects using the CDM is an artificial radio show host that can conduct a full radio program completely autonomously, including interview listeners over the phone.

Constructionist Papers
Cognitive Map Architecture for Honda ASIMO
Constructionist Design Methdology for Interactive Intelligences
From Constructionist to Constructivist A.I.

MINDMAKERS.ORG. The mission of MINDMAKERS.ORG is to further development of large integrated artificial intelligence systems, with a strong focus on real-time human-humanoid interaction and collaboration. The main forum for Constructionist AI. Please consider joining!