Objectives
The objective of building conscious machines was already a research topic in the early years of artificial intelligence, but the extreme difficulties encountered at that time in developing implementable models of even the simplest features of human intelligence halted the research and put machine consciousness into the bin of Utopian research topics (more or less like time-travel, immortality or hair-restoring).
But the case for consciousness is a little bit different because consciousness does exist now. Consequently, we know a priori that the construction of a conscious entity is possible. Research in artificial consciousness is not any longer Utopian research for several reasons:
Recent advances in neuroscience and physiological psychology have provided a deeper knowledge of how human and animal consciousness arises.
Modern robotic mental architectures have refocused from abstract problem solving systems onto systems that continuously interact with the world, which some take to be an important step toward the explanation and creation of (a particular kind of) awareness (i.e. flea-type awareness).
Developments in the field of complex control systems delved into theories of how model-based control-loop integration can lead to an integrated perception of the self of the controller and the plant.
Computer technology now provides computational power that is many orders of magnitude beyond what was available in the past. It has been estimated that in ten to twenty years, computers will reach the computational power of the human brain (from a certain perspetive).
A recent explosion of interest in consciousness on the part of philosophers has led to the most sophisticated conceptual understanding to date of the possibilities for a scientific theory of consciousness, and the potential obstacles to such.
We understand that the main problem for having a good scientific theory of consciousness lies in the wide scattering of knowledge among a wide collection of disciplines. It is necessary to employ coherent, interdisciplinary approaches to the problem to get a glimpse of a good solution. This is why this workshop is essential.
There have been previous attempts to advance the development of a theory of consciousness but, to our knowledge, this is the first attempt to filter theories on the basis of technological applicability (engineering filtering).
Non-implementable theories (i.e. theories that claim that consciousness is indefinable or unknowable, theories that say that consciousness is epiphenomenal and hence has no causal powers, theories that say that consciousness is just a myth invented by philosophers, theories that say that no machine could have it or theories that say that machines that are indistinguishable from us could lack consciousness) are not useful for the engineering work. It is necessary to re-consider their suitability as scientific explanations.
This work is interdisciplinary by nature and by need because there is no single discipline that can provide all the relevant knowledge, nor the necessary broadness, nor the essential tools to build such a machine.
The people who have the data about natural consciousness can be found in the disciplines of psychology, medicine, neurobiology, ethology, linguistics, antrhopology, and philosophy.
The available scientific models of natural consciousness can be found in neuroscience, cognitive science and biophysics.
Some artificial intelligence and robotics researchers now regard artificial consciousness as one of the basic research objectives of the field.
If we are going to “engineer” conscious machines it is obvious that it is necessary to have engineering disciplines represented in the workshop. We have selected people form computer engineering, software engineering and automatic control engineering.
The expected outcome of the workshop is divided into two parts. The first one is a collection of scientific-technical results:
A sound characterization of the field
An engineering definition of consciousness
A seed for a sound theory of consciousness
Draft requierements for architectures for conscious machines of various sorts
The second part is a planning of future research activities in the field:
Research roadmap
Launch a network of excellence
Elaborate concrete collaborative research projects