Field Essay 03 · AI and attention
Can Computation Increase Attention?
A signal does not have to replace perception. It can create a reason to look more closely.
AI is usually asked to make work faster, decisions easier and uncertainty smaller. I became interested in a different possibility: could computation make someone more attentive to a moment they might otherwise miss?
Prediction usually promises closure
A forecast appears to settle a question before experience begins. Rain or no rain. Buy or wait. Act or ignore. The interface gives a signal, and the person converts it into a decision.
This is useful, but it also narrows perception. Once a system tells us what matters, everything outside the signal can become background. The more confidently a system predicts, the easier it is to confuse its answer with the world itself.
An uncertain number can open a window
Pink Sky Predictor was never capable of removing uncertainty. It retrieved live Singapore weather and turned selected conditions into an experimental heuristic probability. The number could not promise a pink sky.
That limitation became part of the proposition. A probability could function less like an answer and more like a prompt: perhaps this evening is worth noticing. Computation would not see the sky on a person's behalf. It would create a small window in which looking became more likely.
Accuracy is not the only measure of value
A forecasting tool should be honest about what it can know. But accuracy alone does not explain what a signal does in a life. A modest probability might change when someone leaves a desk, carries a camera or remembers that the sky is part of the day.
This does not make an incorrect forecast correct. It suggests another design question alongside accuracy: what kind of attention does the interface produce? Does it close observation by issuing a verdict, or does it make the person more available to what actually happens?
The interface between calculation and poetry
I placed a computed number beside a poetic line because neither register was sufficient alone. The number offered structure without certainty. The line returned ambiguity, feeling and the possibility that the moment might matter for reasons the calculation could not represent.
The pairing asked whether computation and interpretation could share one frame without one flattening the other. The desired response was not simply belief in the number. It was curiosity about the sky.
Intelligence can prepare perception
I do not want AI only to finish tasks before I have to encounter them. Sometimes I want it to help create the conditions for encounter.
A useful signal can say: something may be happening; bring your own eyes. The judgement remains with the person, and the world retains its ability to surprise. Computation has not replaced attention. It has prepared attention to arrive.