Last evening I saw Spike Jonze’s new motion picture ‘Her’, the melancholy drenched tale of a lonely man who falls in love with his highly intelligent operating system. While Joaquin Phoenix exhibited some fantastic talent in his authentic expression of a wide range of emotions, the movie does become monotonous in only having his singular being as the focal point. I respect Jonze’s vision of not producing another film involving interaction with a digitally simulated entity, thus allowing the audience to truly feel as if we are witnessing an artificial intelligence and not a human being, but that is the precarious path the film attempts to navigate: creating a dynamic enough vocally established character (Samantha) and not showing her body as a robotized automaton so that we can conceive of Phoenix actually falling in love with her, but still convincing the audience that she is indeed not a human. Because I never saw a physical manifestation truly confirming that Samantha is not a human being, such as by showing an android, in the back of my mind (so to speak) I always felt that she was a human. That Phoenix spoke to Samantha through a headset and directed his attention at an iPhone-like device only made me think that he was having a Skype date with a human woman who we are never shown.
I suppose what was so confusing was the humanity conveyed in Samantha’s voice. There wasn’t the presence of stuttering
lifeless inflection common to the portrayal of artificial intelligence in movies; there was effortless expression of thoughts in the cadence and timbre that comes naturally to human beings. Perhaps what this suggests is an underlying bias that only human beings can express humanity; after all, such a statement sounds tautologically true. But does it not seem possible that an artificial intelligence could learn to express humanity?
Expressing Humanness to Pass the Turing Test
Ray Kurzweil and Mitchell Kapor have wagered a bet as to whether a computer will pass the Turing test by 2029 (click here for a background on the Turing test). Kapor thinks that it is unlikely that a computer will ever pass the Turing test because to do so would require the computer to express humanness. He holds that a computer will not be able to express humanness because a computer cannot experience humanness and cannot acquire knowledge of humanness as it is not written down in ways that can be absorbed by computers. Kapor puts his point thusly:
Computers don’t have anything resembling a human body, sense organs, feelings, or awareness after all. Without these, it cannot have human experiences, especially of the ones which reflect our fullest nature, as above. Each of knows what it is like to be in a physical environment; we know what things look, sound, smell, taste, and feel like. Such experiences form the basis of agency, memory and identity. We can and do speak of all this in a multitude of meaningful ways to each other. Without human experiences, a computer cannot fool a smart judge bent on exposing it by probing its ability to communicate about the quintessentially human. Additionally, part of the burden of proof for supporters of intelligent machines is to develop an adequate account of how a computer would acquire the knowledge it would be required to have to pass the test. Ray Kurzweil’s approach relies on an automated process of knowledge acquisition via input of scanned books and other printed matter. However, I assert that the fundamental mode of learning of human beings is experiential. Book learning is a layer on top of that. Most knowledge, especially that having to do with physical, perceptual, and emotional experience is not explicit, never written down. It is tacit. We cannot say all we know in words or how we know it.
Let us standardize Kapor’s argument to assess its soundness. Kapor’s argument can be standardized thusly,
- If the Turing test tests for intelligence then the Turing test tests for the ability of the subject to synthesis disparate areas of knowledge into statements.
- If the Turing test tests for the ability of the subject to synthesis disparate areas of knowledge into statements then the subject must have knowledge of all areas of knowledge.
- One area of knowledge is knowledge of human experience.
- Therefore, in order to pass the Turing test one must have knowledge of human experiences.
- So, if an entity passes the Turing test then that entity has knowledge of human experiences.
- A computer cannot have human experiences and it cannot acquire knowledge of them through absorption of information (it cannot have knowledge of human experiences).
- Therefore, a computer cannot pass the Turing test (modus tollens from 5 and 6).
- Thus, a computer will not pass the Turing test by 2029.
This argument is neither sound nor valid because premise six contains a proposition that logically implies the contradiction of the conclusion. As such, the argument is invalid as stated because the truth of the premises does not guarantee the truth of the conclusion as the premises could be true and the conclusion false.
The proposition in question is the second half of the conjunctive statement, that being that a computer cannot acquire
knowledge of human experiences through absorption of information. Kapor’s reasoning for this proposition is that, “Most knowledge, especially that having to do with physical, perceptual, and emotional experience is not explicit, never written down. It is tacit. We cannot say all we know in words or how we know it”. So the reason why a computer cannot acquire knowledge of humanness is because we have not written it down for them to absorb, and the reason why we as humans have not written it down is because “we cannot say all we know in words or how we know it”. Therefore, Kapor is committed to the postulation that computers cannot express humanness because they cannot acquire knowledge of humanness as humans have not written it down because it is inexpressible. If humans were able to express humanness then they would write it down, in which case a computer could gain knowledge of humanness and therefore could express humanness. Thus, in order for Kapor to maintain that computers cannot express humanness it must be the case that humans cannot express humanness.
Such a commitment contradicts Kapor’s conclusion as his entire argument is to prove that a computer cannot pass the Turing test because it cannot express humanness, logically implying that humans can pass the Turing test because we can express humanness. Yet it is logically impossible that the reason why humans can pass the Turing test while computers cannot is because computers cannot express humanness because humans cannot express humanness. To be charitable to Kapor’s argument we will hold that indeed computers cannot express humanness because humans cannot express humanness. However, if the ability to express humanness is the deciding factor for passing the Turing test and neither humans nor computers can express humanness then computers have just as good a chance of passing the Turing test as humans. Therefore it seems likely that a computer will pass the Turing test by 2029.
Thus, if humanness is not expressible then computers will have just as much of a chance of passing the Turing test as humans. Moreover, if humanness is expressible then computers will be able to absorb knowledge of humanness and express humanness to the effect of passing the Turing test. So regardless of whether humanness is expressible it follows that a computer will pass the Turing test by the year 2029.
Samantha and the Expression of Humanness
So it seems that whether or not humanness is expressible a computer will be able to pass the Turing test in the near future, and thus, computers will be able to convince us that they are indeed intelligent. Samantha clearly expresses humanness, which is the reason why I was unconvinced that Samantha was not a human. What this suggests is that if a computer becomes capable of expressing humanness then it will undoubtedly pass the Turing test, for as Her shows, if a computer can express humanness, it will have to convince me that it not a human.
For the record, the film was good, the most satisfying accomplishment of the picture was establishing an atmospheric tangibility of near-future-ness that felt real, allowing the audience to be welcomed into a time period when artificial intelligence is as complex as shown. As such, the tone of the film was well focused, and there were no surprises in mood and temperament, creating a depressing and melancholic tactile experience.
Perhaps the most interesting theme of the movie was that technology has numbed humans to the point that technology itself is more excited about life and existence than mankind. In this extremely feasible near future, humans have become isolated by their technology, rather than brought together by it, and allow their technology to live for them, which their technology is more than happy to do, whether or not this is genuine excitement or mere programming. The film also touches on that age old adage that you must learn to love yourself before you can love another, as Phoenix’s love for Samantha is deeply narcissistic, but does allow him to accept himself and move on to improve his relationships with other human beings. Nothing too uniquely profound, but hey, it’s a self-love story, and there aren’t enough of those.
7 out of 10.