Making the simple complicated is commonplace; making the complicated simple, awesomely simple, that's creativity. -- Charles Mingus

Saturday, June 6, 2015

Some Thoughts on Ex Machina

I recently saw the new movie Ex Machina, by writer/director Alex Garland. The movie concerns an inventor (Nathan, played by Oscar Isaac) who invites a minion from his company (Caleb, played by Domnhall Gleeson) for a week at his remote estate. Caleb is to play the role of interrogator in a Turing-test of Nathan's most recent invention Ava.  Nathan wants an independent opinion on whether Ava is an AI.

This post is primarily about philosophy, but if you read on, you should probably expect spoilers.

Ava (played by Alicia Vikander) has a humanoid form- possibly designed by Apple. But Ava is obviously an android - we get to see the machinery within Ava's transparent neck and torso, and she (sometimes, but not always) makes subtle machine-like sounds when she moves.

Early in the movie Nathan asserts that this version of the Turing test, where you can obviously see that the testee is a robot, is much better (harder to pass, we assume) than Turing's formulation (Computing Machinery and Intelligence, A. M. Turing, Mind, Vol. 59, No. 236 (Oct., 1950), pp. 433-460). This proposition is accepted by the interrogator, but I am not so sure.

Turing explicitly sets up the imitation game so that the participants cannot see one another. In the original imitation game, which Turing is repurposing, the goal was to tell which of the participant is a man, and which a woman, so this is perhaps necessary.  But Turing retains this setup. He mentions that this is so that the interrogator is not influenced by "tone of voice", but one assumes that embodiment would be equally concerning.

For me, this is the right decision. We habitually ascribe intelligence to the people that we meet. This is our default response for good reason. By performing the test in the way that the movie assumes, with a walking talking humanoid device, I think that most interrogators would be more, rather than less, inclined to ascribe intelligence to the device.

There are other issues with the philosophy in the movie.  Neither Nathan nor Caleb are quite sure exactly what they are testing for.  At the beginning of the movie, Caleb is quite clear about the Turing test as a test for intelligence.  Later in the movie, both men discuss the test-in-progress, referring at times to consciousness and life.  I'm not sure whether we're supposed to assume at this point that the test for intelligence is passed, and that they are moving on to further considerations.  I think not.  It seems to me that the writing is just confused.

The social politics of the film are quite troubling.  It is not explained why the mute prototype android, Kyoko (played by Sonoya Mizuno), who serves in the kitchen and bedroom, is cast as an Asian woman.  The switch from Asian features to Caucasian in the development of the AI is never explained, and the role of Kyoko just seems to play into the worst stereotypes.  Oh, and the fact that Ava has those servo sounds, and Kyoko does not seems like a regression to me.

No comments:

Post a Comment