This question is less hypothetical than it seems. In recent years, there has been amazing breakthroughs in low-light photography with machine learning, such as [1]. It's not hard to imagine that it would be applied to future surveillance cameras.
I imagine that in most cases, the "AI" would work just fine at guessing things. But important edge cases exist, including the possibility of adversarial machine learning spoofing attacks to create false images. Imagine faking a crime scene this way, it would something straight from cyberpunk fiction.
This question is less hypothetical than it seems. In recent years, there has been amazing breakthroughs in low-light photography with machine learning, such as [1]. It's not hard to imagine that it would be applied to future surveillance cameras.
I imagine that in most cases, the "AI" would work just fine at guessing things. But important edge cases exist, including the possibility of adversarial machine learning spoofing attacks to create false images. Imagine faking a crime scene this way, it would something straight from cyberpunk fiction.
[1] Learning to See in the Dark https://cchen156.github.io/SID.html