Illustration in papyrus and graphite tones, with a polygonal aesthetic. Three stone seats surround a low circular bowl; from each seat, a carved cord extends into the bowl, the surface of which displays intersecting concentric waves.

Public notebook

Before the prompt: genealogies of the artistic interface

Generative artificial intelligence has imposed a short chronology. According to this narrative, art began to engage with technical systems when image and text generators appeared, and immersion, bodily feedback, or agency distributed between person and machine are considered innovations of this decade. The retrospective Mariko Mori: All That Shines, which the Mori Art Museum in Tokyo opened on 31 October 2026, allows for a discussion of this simplification even before evaluating the exhibition.

The exhibition will remain open until 28 March 2027. The museum announces approximately forty works spanning over three decades. The documentation by Alexandra Munroe, co-curator alongside Kataoka Mami, employs a figure close to eighty when including archival materials. These counts are not necessarily incompatible, but they should not be presented as if they measured the same thing. This text is not a review. The exhibition opened on 31 October 2026, the date on which this text was completed. It departs from a work included therein to reconstruct a history of the artistic interface prior to the prompt.

Wave UFO and the fabrication of a first

Wave UFO was created between 1999 and 2002. In 2003, it was installed at 590 Madison Avenue in New York, as part of a Public Art Fund project. Three participants would enter a biomorphic capsule and connect to sensors that recorded brain activity. The information was transformed into projected images and sound within an immersive environment. The work combined biofeedback, computer-generated graphics, and a collective experience of the measured body.

The corporate communication from Mori Building on 23 June 2026 presented it as the first work of art to utilise brainwave biofeedback. The Mori Art Museum website uses a different formulation: it describes it as the largest three-dimensional installation to employ that technique. The second statement limits the comparison to a scale and a typology; the first claims a historical priority that does not hold if read literally.

The difference is not minor. Retrospectives organise a trajectory, but they also produce a history of the medium. When a claim of primacy accompanies a work, it concentrates dispersed antecedents into a single, easily communicable milestone. The problem does not consist in denying the singularity of Wave UFO. It consists in making it dependent on a priority it does not require.

Before 1999

In 1965, Alvin Lucier premiered Music for Solo Performer at the Rose Art Museum at Brandeis University, with John Cage among those participating in the inaugural performance. The technical premise originated in the laboratory: Lucier collaborated with the team of physicist Edmond Dewan, who was studying alpha waves as part of a United States military research programme. The performer remains seated with electrodes attached to the scalp. Alpha waves are amplified and trigger percussion instruments via transducers. The piece inverts the conventional relationship between intention and execution: conscious effort tends to interrupt the signal that produces the sound. One does not interpret a score through controlled gestures; rather, one orchestrates a situation in which the body produces information precisely when it ceases to pursue an immediate result.

From the late 1960s, David Rosenboom conducted extensive research into biofeedback and musical practice. His notebooks from 1969 to 1974 and the volume Biofeedback and the Arts: Results of Early Experiments, published in 1976, demonstrate that this was not an isolated anomaly. Biofeedback had already become a field of experimentation concerning perception, composition, learning, and states of consciousness.

Nina Sobell dates BrainWaveDrawing LA to 1974, although her archive index traces the series back to 1972. Two participants connected to an electroencephalography system observed representations of their activity while interacting. This discrepancy in dates must be preserved rather than artificially resolved: artist archives also contain versions and retrospective decisions.

The institutional framework was equally antecedent. Cybernetic Serendipity, curated by Jasia Reichardt at the Institute of Contemporary Arts in London in 1968, brought together art, music, poetry, machines, and computational processes. It was neither an exhibition of artificial intelligence in the contemporary sense nor a comprehensive genealogy of digital interaction. It does, however, demonstrate that the relationship between calculation, feedback, and aesthetic production already possessed spaces for exhibition and debate.

The genealogy is not limited to electroencephalography. Richard Teitelbaum employed brainwaves as control voltages for a synthesiser. The literature situates these works within the improvisations of Musica Elettronica Viva and dates Spacecraft to 1967, yet I have been unable to confirm the work, year, and use of brain signals within a single open primary source. In 1971, Pierre Henry presented his musical setting of Roger Lafosse's Corticalart. The IRCAM record documents the work and Lafosse's system; that this device converted the composer's brain activity into sound material during the concert is an attributed description that I have been unable to confirm in an open primary source. Furthermore, in 1995, Char Davies presented Osmose, an immersive virtual reality environment in which the visitor navigated through breathing and balance, recorded by a sensor-equipped jacket that tracked respiration and the shifting of the centre of balance. Osmose did not measure brainwaves, yet it anticipated two features that Wave UFO would develop: total immersion and a body that governs the environment without precise voluntary gestures.

With these antecedents, Wave UFO cannot be the first artistic work with cerebral biofeedback without introducing a very narrow definition that corporate communication does not offer. It may remain a decisive installation due to its scale, its combination of collective experience and computational imagery, its architecture, and the integration of that device into Mori's spiritual and technological evolution.

From control to feedback

The prompt appears to offer a transparent unit of intention: a person formulates a request and the system returns an output. That scene favours a simple reading of authorship and control, even though the model's operation, the training data, and the decisions incorporated by the platform remain out of sight.

The interfaces of Lucier, Rosenboom or Sobell propose another relationship. The bodily signal does not formulate a request. The system does not limit itself to executing a verbal order. It measures, amplifies, translates and returns information that modifies the behaviour of the participant. The work is constructed as a circuit, not as a commission. Agency is distributed between will, physiological activity, sensors, amplification, instruments, rules and learning.

This does not automatically convert these works into precursors of generative AI. Calling them such would reduce their specificity to a preparatory function regarding the present. Nor does it establish a moral superiority of bodily feedback over the prompt. It allows for the distinction of two interface models. In one, the satisfaction of an instruction predominates; in the other, exposure to a system response that does not coincide entirely with a request.

The difference helps to formulate a critical question regarding generative creation. The problem is not only how much human control exists or how many iterations precede the result. It also matters what type of relationship the interface permits: whether it reduces interaction to specifying and choosing outputs, or whether it makes its own restrictions, dependencies and misalignments perceptible.

I propose three questions to compare interfaces without hierarchising them by date. The first is what the system measures: a sentence, a gesture, breathing, a physiological signal or usage history. The second, what it returns and with what delay: a closed output that is accepted or discarded, or a continuous response that modifies behaviour while it occurs. The third, who can modify the circuit: whether the participant, the artist or the platform sets the rules of translation between input and output. Applied to a commercial generator, the three answers are usually asymmetric; applied to Music for Solo Performer, the circuit is visible and the performer learns to inhabit it. It is a reading tool of mine, not an established classification.

In terms of the structural surplus, the question is not who presses first, but whether the encounter reorganises something that the intended function does not absorb. Biofeedback interfaces make that question visible because the body does not appear solely as a source of orders. It also appears as material that resists, fluctuates and returns information.

Brain data in 2026

A distinction separates 1965 from 2026: the signal is no longer merely sonic or visual material. Brain activity recorded by a device may be treated as personal data, and capture technology has become a consumer commodity. In October 2021, Chile amended Article 19 of its Constitution to protect brain activity and the information derived therefrom against scientific and technological development. In Spain, the 2021 Charter of Digital Rights, a document lacking normative force, dedicated a section to the use of neurotechnologies.

The implication for a reinstallation of Wave UFO is an inference of my own, not information provided by the museum: if the work once again records signals from real visitors, the institution will have to determine what data it retains, for how long, and with what prior information provided to the participant. In 2003, that question was barely part of the apparatus. Today, it is intrinsic to it.

The interface also ages

Of Wave UFO, we know its dates, several exhibitions and a general description of its operation. The public documentation consulted does not allow for the reconstruction of the number and model of sensors, the software, the system versions, the complete technical equipment, the current ownership or the conservation strategy. That absence does not demonstrate that the information does not exist. It indicates that it is not available to evaluate from the outside how a technological work created a quarter of a century ago has survived.

The imbalance is telling. A claim of primacy circulates in a press release; the technical history that would allow for the comparison of the work, its reinstallation, or a discussion of its transformations remains outside the narrative. The capsule, the cosmological experience, and the image of the connected brains are remembered. Calibration, replaced components, software dependencies, maintenance, and technical labour remain in the background.

Preserving such a work necessitates a decision regarding what constitutes its identity. The original equipment may be preserved while it remains operational; its behaviour may be emulated in a new environment; components may be migrated; or the experience may be reinterpreted using different technology. Each strategy preserves certain properties while modifying others. There is no purely technical answer: one must determine whether the appearance of the device, the mode of participation, the type of signal, the visualisation, the duration, or the historical context of the hardware are essential.

Variable media models and resources such as the Variable Media Network or Matters in Media Art emerged to record these decisions. The exhibition Seeing Double: Emulation in Theory and Practice, presented by the Guggenheim in New York in 2004, displayed original works alongside emulated versions so that the public could compare what was preserved and what was lost. The experiment left a lesson that remains valid: the authenticity of a technological work resides not only in the apparatus, but in the behaviour that the apparatus produced. The same requirement now extends to works created with AI systems: model, version, parameters, input materials, dependencies, limits of variation, and criteria for reinstallation. The history prior to the prompt is not merely a history of ideas and interfaces. It is also a history of obsolescence, substitutions, and uneven documentation.

All That Shines offers the opportunity to observe how Wave UFO is presented today: whether the work functions as it did in 2003, whether it has been reconstructed, what changes are declared, and what information accompanies its reinstallation. Visitors to the gallery may observe four elements:

  • Whether the label or the catalogue distinguishes between original and replaced components.
  • Whether the sensors truly record the visitors' activity or if the projection is a recording.
  • Whether information is provided regarding what happens to the captured data.
  • Whether the gallery acknowledges the technical team that operated the work in 2003 and those who maintain it at present.

This dimension will be as important as the aesthetic recovery of the piece. A technological retrospective does not merely bring together surviving objects. It exhibits, even if it does not always say so, the decisions that made their survival possible.

An open conversation

This text distinguishes documented data from its interpretation. If you visit the exhibition or know of technical documentation on Wave UFO that would clarify its material history, you can write to me.

Sources

Regarding conservation, strategies may be expanded upon in Obsolescence in digital works: migrate, emulate or reinterpret and Technological installations: what a museum preserves when equipment changes.


Discover more from Juan A. Esteban

Subscribe to receive the latest posts by email.

Español English (UK)