Critic discovers painting was human-made. Taste restored immediately.

Art critic Adrian Pell underwent a complete aesthetic recovery on Thursday after a gallery assistant confirmed that the painting he had spent nine minutes condemning as soulless had been made by a person.

The brushwork, previously “algorithmically derivative,” became “a knowing conversation with tradition.” The composition, formerly “the median of a dataset,” acquired “a deliberate refusal of novelty.” A small problem with the hand was upgraded from a telltale artifact to a courageous rejection of anatomical obedience.

The painting remained on the wall throughout. It had not been consulted about any of these developments.

The eye knows, once the ear is told

Pell had arrived at the gallery confident that he could identify machine-made images instantly. This ability, he explained, came from years of close looking and a deep sensitivity to the human condition. Unfortunately, someone beside him had said the word “generated.” He had not waited to hear the rest of the sentence, which concerned the invoice.

Once the error was corrected, Pell returned to the painting for a second look. He discovered emotional vulnerability in exactly the area where he had previously identified a probability distribution. The colors became less suspicious. The horizon acquired a childhood.

His revised review credited the artist with “a haunting meditation on the instability of recognition.” The gallery considered this tactful.

At the neighboring exhibition, a technology evangelist experienced the inverse event. After praising a picture as proof that creative labor was obsolete, she learned it was an oil study from a local painter. She now found it interesting but difficult to scale.

Context is not a crime

There is a serious objection to this little farce. How a work was made can matter. Labor, consent, deception, artistic intention and the economics of commissioning are legitimate subjects. A critic does not owe every object a context-free beauty pageant. An image can be visually compelling and ethically troubling. A handmade work can be dull and fairly paid for.

The problem is not considering provenance. It is pretending that a verdict about provenance was an observation about color, or that a preference arrived through the eyes when it was actually delivered by the press release.

If your objection is that a training practice used work without meaningful consent, make that argument. If your objection is that a client misrepresented the production process, make that argument. These are clearer, more demanding claims than “I can always tell,” a claim that has the unusual property of becoming harder to test whenever its owner gets something wrong.

The same courtesy applies to enthusiasm. A useful generative tool does not establish that every generated image is good. Lowering a production barrier does not remove the need to decide whether the thing deserves to exist on the side of a bus.

The gallery tries an experiment

By Friday the gallery had introduced a new labeling system. Each work displayed its medium, the artist’s description of the process and a separate note about any automation. Viewers were invited to look before reading. Those who wished to discuss ethics could do so without pretending to have discovered a copyright dispute in the negative space.

The experiment produced a remarkable diversity of responses. Some people liked things they had expected to dislike. Others disliked things they had hoped would validate an entire worldview. One visitor simply enjoyed the exhibition and left. The gallery was unable to convert this behavior into a panel discussion.

Pell found the new system troubling. “The uncertainty makes criticism difficult,” he said. Asked whether that was perhaps part of the job, he became interested in a sculpture across the room.

The technology evangelist proposed automatically generating the labels. The gallery assistant proposed lunch. For a brief moment, a concrete human need won.

The final review

Pell’s published assessment awarded the exhibition four stars. His strongest paragraph concerned the urgency of preserving independent aesthetic judgment in an age of technological confusion. It was a good paragraph. The argument deserved to be considered on its merits.

The gallery has declined to disclose which works visitors misidentified. It says the purpose was to encourage attention, not create a leaderboard of embarrassment.

This decision has drawn criticism from both camps, each of which believes the public deserves a transparent account of the other camp being wrong.