Some of the images below were produced by humans and some by artificial intelligence. With major developments in artificial intelligence and style transfer, machine learning programs are now able to discern artistic styles with remarkable accuracy. This information is then utilized to create new works of art or to transpose patterned styles onto previously defined images. Making use of these technologies, the ACT Lab is currently examining how source knowledge (human or artificial intelligence) influences perceptions of creativity, authenticity, and meaning.
Is AI Changing the Way We Think About Authenticity?
July 23, 2017
Could the strategies we use to think about and evaluate art be changing in wake of artificial intelligence? A skeptic might say that artificial intelligence programs could never hope to capture the elusive and distinctly human pursuit of art. Whether that is true or not, it does not mean these technologies won't lead to major shifts in the ways humans perceive art and creativity. To understand this, let's consider a few ways in which we evaluate art.
Economists are fond of saying that art is a "credence good", meaning that it is tough to assign an objective value to works of art. What makes someone pay millions of dollars for a Van Gogh and next to nothing for Van Gogh lookalike? Put more directly, what is art really worth? Are we paying for an object or a story? Economists and psychologists often agree that art is often evaluated in terms of its authenticity. We value art based on knowledge of whether or not it was produced by a specific person, in a certain place, in a special way, or at a momentous time. Indeed, if we know that something was painted by someone famous for instance, we are subsequently inclined to evaluate the work more favorably. Similarly, our research shows that we rate art produced by humans as more skillful and expensive. It could be that we do not perceive an algorithm to possess skill per se, rather skill arises from practice and overcoming failure. No matter the explanation, the importance of source information may be referred to as index authenticity and it is but one strategy employed in the evaluation of art.
This is not the only form of authenticity we use to evaluate art, however. Another form of authenticity can be called "artistic authenticity" or "motivational authenticity". This refers to when a work of art was or was not produced for some external reason. Was the artwork commissioned, was it produced for reward or merely to reflect something internal? When a music artist appears to be solely motivated by money, for instance, their work is often subsequently devalued, and their artistic authenticity questioned. Instead of focusing on the "who" or the "how" this strategy of motivational authenticity places focus on the "why". Why was this art made? Was the artist producing work because they desired fame or because they desired pure self-expression and creative freedom? The latter is often praised, the former disparaged.
Complicating this, we might break motivational authenticity into two distinct parts. One part might be said to involve the absence of resistance to extrinsic motivation. Was the art influenced by the pursuit of fame, money, or social pressure? The other part might be said to be the presence of intrinsic motivation. Was this motivated by the need for self-expression or emotional catharsis?
This division may seem arbitrary at first glance but may take on increasing importance as technology continues to develop. AI programs occupy a particularly tricky space when considering intrinsic and extrinsic motivation. On the one hand, one might argue that AI programs lack the necessary emotional expressiveness, the internal, intrinsic drive to produce art. On the hand, one could make a similarly strong argument that art produced by AI is never produced for the all too human and extrinsic motivations of money, fame, or respect. They were made to produce for no other reason than production, a narrative often considered laudable in the human artist. After all, what human has ever produced art without at least some small desire for personal gain or recognition?
The introduction of AI in the art world should allow us to look at these different kinds of artistic authenticity in a new light. As such, our lab has begun to measure whether or not ideas of humanity and artificial intelligence influence our perception of art. In particular, where and how do we assign meaning to art. Does this meaning influence evaluative price? Without knowledge of the source, do we see work produced by humans and AI differently? When the source is made salient do our perceptions change? Can this be complicated by priming artistic ideas around intrinsic or extrinsic motivations? Stay tuned for exciting news and research updates!