On Emotion in the Creative Process

I have written a lot in this blog about AI and machine learning, and have probably established my credentials as a sceptic. It is not 'intelligence' and it will not, in my view, ever be able to create anything other than curious pastiche. That is not the same as saying it will not create things that people will utilise: most popular entertainment is basically pastiche. And these days 'entertainment' seems increasingly to mean addictive clickbait, at which AI will presumably excel. But it will not produce anything really worth reading, watching, or hearing.

This is because - I know this will shock and appall readers - human beings are not rational. We make our decisions on the basis of emotions. And anything that does not have emotions therefore cannot replicate human thought or decision-making. In this regard, I strongly recommend listening to this interview with Robert Burton, a neurologist who has written extensively on knowledge and decision-making. A transcript of the crucial passage in the interview runs as follows:

[There was] a cardiac surgeon of some repute, who did a study of whether or not hands-off massage--I have forgotten the name for it now but it's when you run your hand over the patient's body but don't actually touch them--will improve cardiac surgery. And, when they asked him why he came up with this idea, he said, 'Well, I had no a priori opinion on this.' Then, I would say, 'Why would you do this study?' I mean, that would be the equivalent of saying...eating lasagna helped cardiac surgery. You'd say, 'Why?--' and this was sort the plea that I have in my second book, is that: Scientists initiate almost all research, and I mean, I say, 'almost all' I'm just trying to be generous, from the point of view of some preconception. Often one that they don't understand at all. But it's just one that tweaks them. And I was--you think about Albert Einstein and the theories of relativity, and he was working at the Swiss patent office, and one of the big issues at the time was with the nature of time and getting railroad scheduling. So, trying to arrive on time. And he wasn't the only one thinking about it. Now, the question is: If he hadn't worked in the patent office, would he have come up with the same idea? Maybe. Maybe not. But did thinking about time and getting it so the trains--triggered an experiment about the man on the train? Well, you never know. I wouldn't call that a bias. I would just call that prior experience and his native temperament have shaded the way he starts thinking about the experiment. And that's not overcome-able.

The crucial phrases are there in bold. Human beings - even scientists, who are purportedly 'rational' - decide what they are going to investigate, and how they are going to investigate it, on the basis of emotion. Otherwise, why would they be interested in the thing they are investigating, as opposed to the infinite range of other topics they could be investigating? Why are they interested in investigating anything at all? The answers to those questions are based in 'prior experience and native temperament', on how the scientist is being 'tweaked' (which is to say, irrational motives), and not on reason. Why did Einstein investigate relativity? Because he was interested in it. And the 'being interestedness' is itself rooted in emotion, not rationality.

The emotion, then, comes first and is crucial. Why do we get out of bed in the morning? Because we feel that there is a reason to, as opposed to not doing it. More pertinently, why do we want to create a piece of art in the first place, let alone actually go about the process of doing it? Precisely because, well, we want to. The feeling and wanting are necessary - they are where volition comes from. And they are what dictate to us the direction in which we will go during the creative process. Human creativity is in other words only partly iterative, and only partly based on prior influences and knowledge of the genre in which one is working. It is emotion that dictates the decision-making processes which are continually made during the production of any given work of art.

There is a great interview available on YouTube between Rick Beato and Billy Corgan. You'll get a huge kick out of it if you're a Smashing Pumpkins fan and should listen to the whole thing. But if you're not, and just want to get to the salient segment, start listening at this point, an hour and five minutes in. Having been asked about the songwriting process, Billy makes clear that drawing from existing influences is only a very minor part of the exercise. Again, to provide a (somewhat paraphrased) transcript:

'Sometimes it might help when you get stuck to think...well, what would John Lennon do? What would Bob Marley do here? Sometimes that can just get you across the line... But as far as the core of what I do, it's always a mystery to me. And the best way I can describe it to somebody is, I'm at home playing the piano, and I'm singing a melody, and I'll sing a note, and I'll think, that's kind of a weird note, so I'll find the note on the piano... [and sometimes] it's dissonant....and I'll think, 'well that's wrong'. So I'll resing the melody, 'correctly', and there's a little guy in my head that goes, 'No', and there's an argument in my brain, and I cannot, not hear the melody that my brain is telling me to sing, so that's the melody.... and if I derogate from it, there's a voice in my head that says, 'No, that's the wrong melody'...'

He goes on (very illuminatingly when thinking about AI):

'There's the computer part [in my brain], and then there's the part which is felt emotion. It's hard to explain.... You're playing something and you think, well, this part's okay. You try something and you go, well that's a little bit better. But maybe it's too weird or out of context...and then you're into the binary choice of whether to go for the D, which is the 5th, or I could go to the F which is the flatted 7th of a G or something, and then you sing one way, and then you sing the other, and you sit there and go, eeny meeny miny mo....I think that's the moment that makes you a songwriter.'

The point, of course, is that the 'eeny meeny miny mo' moment is where emotion comes in. At any given point in time, when writing a melody, one could come up with one note, or another note, or indeed any other of a range of notes. At that level, the human creative process is the same as an AI process. The difference is that the decision of which note comes next to the human is rooted in emotion, whereas to the AI it can only be rooted in reference to other songs which it 'knows'. The human can make a leap based in the logic of feeling. An AI can only, metaphorically, act randomly or by reference to 'What would John Lennon/Bob Marley do?' reasoning. 

This is the difference between art and pastiche. And this is why 'artificial' 'intelligence' will not produce art.