In this piece, John Searle critiques the notions of machine consciousness presented by Nick Bostrom and Alan Turing by questioning their assumptions about what it means for a computer to “know” something. He presents a distinction between observer-independent reality (which exists regardless of what we think) and observer-relative reality and claims that while human consciousness is observer-independent, machine consciousness is not. An example he uses to illustrate this point inolves a machine performing a simple computation. Though a computer can add 2 and 2 together and output 4, Searle argues that this output is completely insignificant without human interpretation. Thus, a computer can't really compute anything alone.

Amidst the growing number of works that focus on the seemingly inevitable rise of super-human machine intelligence, this piece provides a literal “reality-check” regarding the multiple layers of abstraction that exist between a computer’s interface (i.e., how it might response to your inputs) and its physical computing hardware. While Turing equated machine intelligence with it’s observable behavior, Searle reminds us that intelligence and cognitive processes are closely, perhaps inextricably, linked. Moreover, he pushes readers to acknowledge the role human biology plays in the creation of human-level intelligence.

-- Megan Woodruff

Image and Article: http://www.nybooks.com/articles/2014/10/09/what-your-computer-cant-know/...