ME / RESEARCH TOPIC

Software Verification

2 public research records in this topic.

Working paper — not peer reviewed

Beyond the Green Checkmark: Trusting Software That Works While We Sleep

Evidence, limits, and human control in autonomous software delivery

Software can continue working while a person is absent, but passing tests cannot prove that it understood the full goal. This paper asks how such a system can make useful progress without claiming more certainty or authority than its evidence supports. It calls that boundary the verification horizon and proposes practical controls around it. The model is a personal engineering inquiry, not proof that autonomous delivery is safe.

Research note in development

How Would We Test a Digital Mind Without Fooling Ourselves?

Candidate checks for perception, memory, identity, judgment, and longitudinal behavior

A system that imitates intelligent behavior can appear more capable than it is. This developing research note asks how a cross-disciplinary test program might challenge claims about a cognitive twin. The proposed checks are a research agenda, not proof of consciousness or human equivalence.