It also helps to know the whole story.
Lazar J. Greenfield
Professor emeritus of surgery
University of Michigan at Ann Arbor
Christof Koch and Giulio Tononi [“A Test for Consciousness”] defined an experimental method that seems likely to improve significantly on the Turing test as a way to operationally define and identify “intelligence.” The use of sensible versus nonsensical composite images would surely pose challenges to machines—today and for the foreseeable future. I think the article has two weaknesses, however. First, I think the authors underestimate the rate of progress that artificial intelligence will make in this area if it is deemed important. As they point out, the human ability to discern implausible relationships is based on a vast amount of knowledge acquired from experience. The foundations for giving machines that experience have been under development for decades and are gaining traction in many application areas today. It would be silly to take on faith that these tasks are fundamentally or nearly beyond what machines can do.
The second weakness, in my opinion, is that it confuses consciousness with integrated knowledge. Requiring machines to demonstrate that they understand visual elements and relationships seems a straightforward and appropriate aspect of intelligence. But it does not mean that any machine that exhibited that kind of perceptual and cognitive capability would obviously be conscious.
At its core, consciousness is a term we use to refer to our common human perception that we exist, are aware of ourselves and are aware of our being part of the environment with which we are interacting. Self-awareness and awareness of self versus our environment would seem to be important attributes of consciousness, regardless of how it might ultimately be defined and identified. The authors’ proposed test neither depends on those attributes nor distinguishes those having them from those lacking them.
Professor of information systems
Naval Postgraduate School
OUR QUANTUM WORLD
In “Living in a Quantum World,” Vlatko Vedral insists that “quantum mechanics is not just about teeny particles. It applies to things of all sizes: birds, plants, maybe even people.” But all his examples of entanglement refer to the teeny particles—atoms and molecules. The fact that, in some examples, the entangled particles are located within organisms—birds, plants—does not prove that these organisms themselves are entangled. Do the particles and the bodies behave according to the laws of quantum mechanics? Vedral’s answer is affirmative. But that something appears that way to the author and his colleagues is not a sufficient base for sweeping generalizations.