Samar Ansari

Senior Lecturer in Computer and Engineering Sciences, University of Chester. Founding Programme Leader, MSc Management of Information Systems and Digital Innovation.

Samar Ansari

I treat AI as a systemic risk object, and I audit the claims made about it.

The test I apply is independent reconstructability: can a claim, an output, or a system behaviour be reconstructed, verified, attributed, and causally traced, rather than simply judged acceptable? I call the method epistemic auditing.

More on the research programme

Three lines of audit

Cost

Whether AI environmental and compute cost claims are measured consistently enough to be compared.

Infrastructure

Whether the training, inference, and compute stack can be governed in practice.

Knowledge

Whether the evidence base institutions rely on holds up under checking.

Position

Elsewhere

Scholar record: 4,892 citations and h-index 26, as measured on 18 August 2026. GitHub hosts PrivHAR-Bench, an open privacy benchmark for video activity recognition.