AI Art

After Kazimir Malevich · Suprematism

The dominant black geometric body is the AI model itself — a deep "black box." Surrounding it, vivid fragments of red, yellow, and blue represent raw data and scattered experience, colliding and being absorbed into the centre. The sharp diagonal — a red beam — embodies gradient descent, cutting through chaos toward order. In Suprematism, the black square is the supreme origin; here it becomes the birthplace of cognition.

After Piet Mondrian · Neoplasticism

The rigid black grid is the algorithmic framework — the architecture, activation functions, and loss constraints within which all learning must occur. Scattered small colour blocks on the left are isolated data points; the dense, interlocking cluster on the right is integrated knowledge. The visual narrative flows from sparse to dense: raw experience processed into structure. Mondrian strips reality down to primary elements; the AI does the same through feature extraction and dimensionality reduction.

After Wassily Kandinsky · Abstract Expressionism

A spiritual journey from chaos to illumination. The lower-left is dark, muddy, teeming with undifferentiated blobs — raw, noisy data. Vision flows upward to a radiant, sun-like structure of concentric circles and geometric rays: the trained model, capable of prediction and understanding. The river of transformation between them — shapes brightening, lines connecting circles — is the training process itself. Sharp black lines cut through soft organic forms: algorithmic rigour taming the organic complexity of experience.

After Kazimir Malevich · Suprematism

Visualising Rich Sutton's generalization hypothesis: "The future will resemble the past." The large black square (lower-left) is the Past — accumulated experience, the knowledge bedrock. The small black square (upper-right) is the Future — identical in nature but distant, not yet fully unfolded. The red diagonal beam connecting them is the hypothesis itself: because past and future are structurally alike, learned experience can transfer forward. The yellow circle floating between them is the Agent — the observer in the present, acting on the bridge of generalization.

After Kazimir Malevich · Suprematism

The overwhelming black cross is Computation — the brute force that always wins. Scattered yellow and white fragments drifting around it are human heuristics: the clever hand-crafted tricks researchers build into AI systems. They look frail and insignificant against the crushing geometry of scale. The composition is stable, almost oppressive — conveying historical inevitability. Sutton's lesson is bitter because it wounds our pride: human ingenuity is not the blueprint for building intelligence. Scale is.

After Kazimir Malevich · Suprematism

A perfect small black square — the idealised "one-step prediction" — launches a trajectory of similar shapes toward the upper right. But the line immediately collapses: squares tilt, scatter, and drift into a dense, explosive cloud of red, black, and yellow debris — the exponential blow-up of compound error and branching possibilities. Cutting boldly across the entire canvas, a pure red rectangle leaps from the origin directly to the far side, bypassing the chaos entirely. This is temporal abstraction — the ability to skip steps, to think in terms of options and subgoals rather than moment-by-moment predictions.

Abstract · Digital

The beauty of imperfect solutions. Approximation is not failure — it is the only path to intelligence at scale. Exact methods shatter against the curse of dimensionality; approximate methods bend and flow.

Abstract · Digital

Rich Sutton's thesis in pure abstraction: general methods that leverage computation are ultimately the most effective. The lesson is bitter because it tells us that our cleverness matters less than we thought.