Most people have never chosen to use it, and most people are already looking at it. AI upscaling now sits between the content and the screen on phones, televisions, consoles and streaming boxes.
A quiet default
The appeal for manufacturers is obvious: a cheaper panel or a weaker chip can present a sharper image than it has any right to. The appeal for viewers is that a low-resolution stream stops looking soft. Neither party has much reason to advertise that the extra detail was reconstructed rather than recorded.
Where it stops being cosmetic
For entertainment this is a fair trade. It becomes a different question when the image is evidence. A reconstructed number plate, a sharpened face in a security still, a clarified line of handwriting — these are inferences presented with the confidence of a photograph, and AI upscaling is very good at looking certain.
What is reasonable to ask for
Not a ban, and not alarm. Two modest things: that devices say when processing of this kind is on, and that anyone treating an image as a record can switch it off. Both are achievable today, and neither costs the viewer anything.
The technology is genuinely useful. The problem is that it is invisible, and invisible defaults are the ones nobody gets to consent to.
About the Author
