'Your Go'. A thought experiment based upon counter-factual narrative: Between 9th-15th March 2016 AlphaGo (developed by Alphabet Inc.'s Google DeepMind) played Lee Sedol, one of the best players at Go ranked 9-dan, at the Four Seasons Hotel in Seoul, South Korea. In the 5th and final match Lee Sedol, realising he will be unequivocally defeated by a machine, publicly exhibits the common human condition of 'sore loser syndrome'. With a single swipe of his forearm Sedol casts the remaining stones off of the board, scattering them onto the stage to the complete shock of audience, commentators and organisers alike. What is AlphaGo's next move?
The current advance of Artificial Intelligence is now commonly tested within gaming environments. These environments both allow for a relatable context to outside observers and multiple possibility, yet predictable, landscapes within which to apply machine learning. The interrelationship between a (conceptually) hermetically sealed environment and the possibilities for the application of machine driven agency is often neglected by both the creators of AI technology and mainstream press however. In this case the public was continually informed that "despite its relatively simple rules Go possesses more possibilities than the total number of atoms in the visible universe". Such a statement, arguably, leads us to wrongly consider Go as being as complicated as the reality it is contained within. With this is in mind we need to increasingly counter act our own tendency to project human like agency onto artifice. Which brings us to the above thought experiment. By breaking the hermetic nature of the conditions of this event (/PR stunt to advance public acceptance into the mainstream application of AI) by introducing unpredictable traits of human behaviour and human culture we highlight the hard limits of artifically created intelligence and machine learning. Given the event of sore loser syndrome what can we consider to be AlphaGo's next move or options? To assume a malfunction with input data? To issue an open apology to it's opposition and invite a rematch? To enter an eternal pause in waiting for the game / input data to resume? To acknowledge the game as forfeited having being preprogrammed for such an event?
Only by highlighting and contemplating the complex impasses and immunities of coupling machine competence with the unpredictability of the human condition / the chaos observed in natural systems can we confidently advance the given remit for reason based decision making we increasingly assign to machines.
email@example.com | @_ted_hunt |