Hope and Reality in Artificial Intelligence (Marcio Rocha); This is not a Test: Undecidability In-between Machines (Robert Jackson)

This seminar will be a joint two part presentation by Marcio Rocha and Robert Jackson. It will open up a discussion of the interrelationship between developing understandings of HCI and Artificial Intelligence. It will invite basic question about the location and conceptualisation of intelligence and stimulate the consideration of intelligence as an absolute quality that is subject to historical change. The seminar will draw these questions into  some basic debates that inform technological approaches to the intimate involvement of humans with machines.

Rocha-Jackson

Marcio Rocha will depart from an Anthology of HAL-9000, the computer in the science fiction film 2001: A Space Odyssey (1968) to trace the various phases of AI in order to stimulate some reflections through a critical and philosophical view to inform  the design of intelligent machines, addressing some major issues in the field of artificial intelligence.

Robert Jackson will examine an alternative reading of the ‘Imitation Game’ or ‘Turing Test’, first devised by Alan Turing in the NPL report Intelligent Machinery (1948) and in the famous text Computing Machinery and Intelligence (1950). Contrary to many biographical interpretations (Hodges 1992; Leavitt 2007; Copeland 2004), this reading bridges Turing’s original negative solution to the Decision Problem in On Computable Numbers, with an Application to the Entscheidungsproblem (1936) with his later work on machinic intelligence.

In doing so, the presentation will propose that Turing’s proof of undecidability in theoretical basis of Turing Machines provided a direct insight into a speculative understanding of “uncomputable intelligence” within real-world computers. From its inception, universal digital computation is stymied by undecidability; a term used to describe the mathematical predicament where no general systematic procedure can be executed to decide, or deduce what the outcome of another systematic procedure will be.

This reading underscores not just how structural relationships in computing mediate human use (and vice-versa), it also explains how formal computational systems are never holistically connected and are constantly prone to malfunctions, exploits and errors. No sophisticated computational system, whether it is human intelligence, servers, operating systems or smart-phones, can ever fully decide on the output of another complex system.