Up until very recently, we have understood creativity to be the sole preserve of human beings. Over the last twenty years, however, we have seen signs that machines are catching up fast, and the range of creative machine activities is about to explode. Either we will harness the energy of that explosion, or (at best) be left in the dust as AI leaves us behind.
Scientists have been working with automated discovery of new mathematical theorems and scientific hypotheses for over twenty years, now. Naturally then, some artists have wondered what would happen if the same ideas were applied to the arts and culture, and put their ideas to the test in various ways. Fast forward to our current world of predictive advertising, hologram artists and procedurally generated game environments, and we can see that the “low hanging fruit” of the automated arts idea have been picked. Still, however, there is much more that could be done.
What if, as our ancestors intuited, there are sublime artistic and/or spiritual truths which humans simply cannot fully grasp? And what if machines are capable of exploring beyond the boundaries of human limitation in our search for the sublime? In other words, what good things could a fully developed “automated arts” paradigm do for us as a society, and how might we go about creating such a paradigm?
We all hate awful spam-bot adverts and so on, so why would we want more of them? What is it about automated creative content that could be so good that we are ready to inflict an inevitable torrent of automated dreck upon ourselves? Well, aside from the fact that someone is going to do it anyway, the simple answer is that – just as we’ve seen with automated mathematics and science – AI will almost certainly be able to make connections that we simply can’t see. It will open doors into new realms of inquiry and experience for us, if we let it.
To put that another way, advanced AI will be capable of so much more than simply “joining the dots” in content creation, like some digital hack. It will be able to truly explore, and make clear connections that simply had not occurred to us, as a society. That kind of power is invaluable in science and mathematics, but not only in those fields. Powerful insights could serve us very well in many fields, from engineering to the arts, economics to politics.
So if it should be done (and will be done), we should still ask: What might a full-blown version of automated content creation look like? Of course there are many ways in which things might play out. The future is a strange and complicated place, after all. But here is one possibility:
Data Mining is the extraction of patterns from large datasets, to create smaller, abstracted data sets which can be analysed for practical purposes. It is, in essence, a process of simplication that makes incredibly complex datascapes tractable, and useable by humans. For historical reasons, scientific and technical fields have tended to be more quantified than the arts or culture in general – meaning that early AI has naturally been best suited to sifting through scientific data to look for interesting patterns. But as AI itself becomes more sophisticated and more “nebulous” fields (such as social interaction) are increasingly quantified on a massive scale, it seems quite reasonable that AI will soon be able to see patterns in such things that we ourselves are unaware of.
Imagine a kind of Culture Mining, which is to say a paradigm in which algorithms extract patterns from complex artistic, cultural, and social datasets. More than simply doing a fancy Factor Analysis (i.e. looking for relatively independent factors which explain the complexity in a dataset), such methods could go several steps further and apply a kind of “Fractal Dialectic” method, which is to say:
– look for interesting patterns,
– then posit the ideas, artworks or whatever that might bridge gaps between them,
– then look again for patterns between those,
– iteratively rinse-and-repeat,
– and see what strange new forms emerge from the brew.
Interestingly enough, something like that idea was conceptualized by author Hermann Hesse, in his 1943 novel Das Glasperlenspiel (The Glass Bead Game). In the novel, society’s most precious ideas and cultural icons were recombined in a kind of mathematical game, the playing of which was a matter of intelligent governance as much as high culture. It would seem that such a game could in reality only be played by advanced Artificial Intelligences, but who knows… perhaps Herr Hesse was on to something.