Desire for Connection
To reach her goal, Van Alphen applies a stylistic form she
calls roborotics. Roborotics is infused with synthetic sensuality and revolves
around the desire for connection between human and machine, a recurring theme
in her work. With dancing robots and AI-generated bodies, Van Alphen wants to
seduce us into questioning the fusion of technology and humanity, by embracing
it. For this she seeks a voluptuous and playful aesthetic, something we do not
usually associate with technology. And perhaps rightly so.
Sensuality as an ultimate human experience seems at odds
with technology. In her search for an imagery of roborotics, Van Alphen is
actually thwarted by the techniques she works with: the built-in censorship in
tools like Runway ML, Stable Diffusion, or Midjourney makes it difficult to
create original and sensual images. Because of the specific, often limited
aesthetic preferences of AI image generators, every output quickly risks
becoming cliché, slick, and standardized, or, in the case of Stable Diffusion,
oversexualized. This makes it impossible to convey something original or
authentically sensual.
Van Alphen: “What is truly human quickly disappears with
AI, falling between the cracks of idealization and censorship. I’m searching
for something intimate, something sensual. The rawness and softness that are so
inherently human, that’s what I’m after. I try to achieve that by working with
and against the technology.”
Prompting
To nonetheless gain control over AI’s output, various
strategies are possible. The most important act to generate anything at all in
AI – but especially to move past built-in bias – is prompting. Prompting
means giving an instruction to achieve a specific result. A prompt can be an
image, a question, an assignment, a description of a feeling or a movement, or
even a stylistic request. Based on this, the AI delivers its output; the image
(or, if desired, a text). After the machine’s first response, the result can be
endlessly refined with new prompts, until the desired outcome is reached.
The series This is as naked as I could get is a kind
of spin-off from The Oracle, focusing on the search for “the most naked
image” the machine can generate: somewhere between the oversexualized and the
super-prudish, a world that no longer seems to exist in the digital domain. The
visual prompt strategy she used here to circumvent the algorithms of AI image
programmes was the use of ‘fictional’ or quasi-mechanical replications of
motion blur – an analogue technique in photography –
combined with digital textures. Generative AI (still?) struggles with
ambiguity, physical distortion, and analog errors. By giving prompts that are
difficult to interpret, a kind of confusion arises in the machine, allowing the
AI to move past the average aesthetic norm. By reinforcing this through new
prompts, and—as Van Alphen does—offering input in increasingly complex and
formal visual language, you can bypass bias further and further.
The most common form of prompting, however, is in language.
A good prompt can be used as an artistic tool, and must provide direction
without being too deterministic. It can encourage the AI to produce something
surprising, leading to an interesting exchange between artist and machine. A
prompt points: that way, something like this, but better.
