[Self] Confidence Reloaded

[Self] Confidence Reloaded (2022) Early Iterations of future exhibition pieces

Short Description [Self] Confidence Reloaded is an installation where the audience uniquely feels the tension between the dangers of being and not being surveilled. The proposal consists of two wall-mounted sculptures, the original [Selt] Confidence (2019, see www.studioalight.com/SelfConfidence) and a second iteration which is proposed to be produced for an exhibition opportunity yet to be determined. The Installation tells. In confidence. and only directional audible to one individual at a time, the secrets that are usually kept from us.

Using EdgeAl technology, both sculptures map the room locating a visitor. Once located, one or both pivots their "head" and announces directly to the visitor what it observes e.g. "fur coat" plus it will score how confident it is in its own observation, e.g. "42%". The sculptures expose the action of surveillance as aggregated information that a machine is able to perceive. "to make sense of" and reports the surveilled "result." The sculptures inform the assumptions they decipher.

The two-headedness of this installation creates a discursive space, they each reveal their biases and discrepancies [multi-scalar] as they are built three years apart and are programmed with for example different training sets. This dialogical space exposes that both machines have conflicting opinions. By exposing this conflict the goal is to produce a nuance retection within each visitor on the many powers and interests at play.

Contextualization [Self] Confidence Reloaded intends to create a dialogue and exposure of scale, particularly when scale monitors, measures, questions and constructs our perceptions, emotions, and, even, democratic and technological processes and infrastructures. The work not only monitors and measures but it gives, in confidence, a chance for the surveilled to recognize the experience of being surveilled. And, it does so while keeping the beautifully ironic yet complex emotions for those involved in the surveilling. This, in turn, calls any audience member - particularly participants of democratic processes – to participate in our technological infrastructures. The installation invokes a realization that we all must be aware and active to counteract the passiveness of those in power and the “Surveillance Capitalists” (Zuboff, 2014) assume the general public will always be. [S]C Reloaded, calls the audience to pay attention to how power interacts with processes of mapping, syncing, calibrating, and prediction. [S]C Reloaded employs algorithmic processing to shift an audience’s understanding of what is happening while being surveilled. The installation operates via programming and EdgeAI, mapping, syncing, calibrating, and predicting, and because of this it affords embodied contemplations of power. The work includes object recognition and image classification while also uttering the machine's “confidence score”. The installation also aims to provide an exhibition experience designed to reveal that we each are equally small yet connected within “hyperobjects” (Mortan, 2013). It addresses scale, particularly scale in terms of shifting perceptions, in multi-scalar ways. One of which is that each visitor becomes individualized, under the gaze of the works, the head of each sculpture pivots and settles its aim at one person at a time. Once the target is found, each then exposes the bigger societal and technological picture and for example, they reveal their training [equally smart and ignorant] and expose how confident they each feel after being trained. A two-headed installation is proposed for the next iteration of the [Self] Confidence series. This forms a discursive space where each sculpture bears the biases and discrepancies in its individual build and where audience members can be stuck in the middle between two conflicting machine-learned viewpoints. Yet, also, within a crowd of visitors, the machines will most likely point their heads at different subjects at a time, and thus then, the group as a unit can relay not only a machine’s biases but their own through the comparison of the readings across the crowd. An art critic wrote that the original one-headed sculpture (2019) announced that it was “confident” that it was seeing a “rubbish bin” while pointing in their direction. The critic relayed the sentiment of many audience members, it is confusing, and even comical, to be told that the machine is “34% confident” that it sees a “rubbish bin” after it has wobbled its head and pointed its attention at you in the gallery (see Arvidsson full quote in the next section). We built this first version of this piece to be purposefully ironic, and during the many months that the first version was previously exhibited, there were discussions on what it felt like to be “confidently” surveilled. Yet, also, there were discussions on the hidden labor of classification systems, and understanding that surveillance happens at all times by machines. This artwork discloses many levels of labor from the moment of interaction, to our reflections on how the system came to the labels of its classifications. Moreover, [S]C Reloaded highlights an AI’s precondition of having an eagerness to please; it does not know that it does not know. Instead, it mislabels and does so with a percentage of confidence. Often, it is ridiculously confident. While, on the flip side of this ridiculousness, the work airs the fact that autonomous machines open up to the infinite identities between 0 and 1. That between 0 and 1 is fluid with possibilities of classifications and recognitions which could [or does–when meeting their fullest potential] go beyond the binary. All AI really does is increase humanity’s attention to detail, and it does so at scale. This installation, built with EdgeAI, affords a discussion on the act of surveillance and the shifting capabilities of this industry. Surveillance is now aggregated information that a machine is able to perceive, “to make sense of,” and report its surveilled “result”. Each [including our sculptures] reports the assumptions the machine deciphers. In the field of surveillance, the scale of sensemaking enacted brings into question the powers behind it. Which technology makes the most sense when there needs to be a diversity of sense-making at hand? This installation will uncover and create a discussion space around this and more. Format and intended audience In this piece, Studio Alight, proposes to build two sculptures in dialog with each other and the audience (see proposal illustration). These two sculptures, making up [Self] Confidence Reloaded, will run simultaneously and across the same exhibition space whereas each sculpture is wall-mounted and beaming out its particular confidence from two different spots. The second of the two sculptures will function similarly to the first while the contents will reflect the state of technology [including biases and classifications] available in 2022. Thus, the audience can experience the change between these two points in time, 2019 and 2022, for example, in 2019 it classified a bearded person often as a “ski mask” while today it is likely classifys the same as an “N95 face mask”. These three years in our history while battling a global pandemic is where societal shifts are extortionate and where there is a collision of both new habits, e.g. seemingly more open conversations on battling white supremacy, and old habits, e.g. falling deeper into political polarizations spurred on by the isolation of our algorithmically created social media bubbles, or even, Amazon’s unconscionable financial success during this period. The proposed [Self] Confidence Reloaded will create an experience for the audience to realize and ponder the complexities of these technologies in use and we will do so where the installation aims for the audience to realize that being passive is not good enough.

In The Age of Surveillance Capitalism (2014), Shoshana Zuboff postulates that the goal of Surveillance Capitalists is to change us. They capitalize on moments when we connect with our families, while searching our media sleeplessly, and even mundane everyday activities. This is a far cry from the exploitation of labor. Thus, what is at stake is our agency in how we behave, choose, connect, vote, and more. [Self] Confidence Reloaded tells us, in confidence, what it has seen. In turn, the audience is provided with the space to reflect with an individualized –emotional– response to the limitations of the systems that surveil us. Hence, this artwork creates an experiential understanding of how overwhelming surveillance capitalism is impacting so many of us at every moment of every day. Yet, the work does so in a way that one can have an individualized revelation of this impact. Described simply, the format of the proposed work is two interactive sculptures that map the space, and, then, use one neural network to identify humans with a “confidence score” and a second neural network to classify that human and their surroundings. The installation consists of directional speakers, stereo cameras, servos, Edge GPU, trained neural networks, metalwork, and other mixed media. Technologically, this work comes alive using code as an artistic medium, particularly algorithms and machine learning. The stereo camera maps the space in real-time. The algorithms used in the first iteration (2019) are an object recognition algorithm, YOLO, and an image classification algorithm, Googlenet, trained on the ImageNet [and before this dataset was deemed unethical for use]. The 2022 sculpture will be a new iteration. An art critic wrote “[...] Studio Alight plays with the intrusive surveillance community by incorporating an image recognition system. I do not know how to react when a voice belonging to the work tells me and other people who visit the exhibition that we are rubbish bins, but maybe it is better to be humiliated than not get any response at all? In the digitalized world, the height of shame is not to provoke any reaction at all. [...] Behind the ironic joke is a painful reality.” (Translated from Swedish by author, original Sara Arvidsson, Göteborgs Post, 31 Aug. 2019).” The original piece was robustly tested and shown in the Rob Law Exhibition at ICIA in Gothenburg, Sweden from August to December 2019. This work discloses the secrets that are usually kept from us in the seemingly alternative universe. The sound enters the space via a directional wave; thus, the others in the room cannot hear the classifying when it is not pointed directly at them. Also, these two machines are overly concerned with revealing their measured confidence. It speaks out its confidence score, confessing how confident it reads the image, or how confident the machine distinguishes that it has chosen a classification. This postulates that the notion and application of “confidence scores” guiding autonomous machines will [undoubtedly] become an evidential element in a future court case, e.g. how confident do autonomous armed drones need to be? And if this question is flipped, just how confident are we, humans, that our experiences are being handled ethically by those in power?