New hazard idea: bugs/reproducibility/error

[NatalieZelenka]

A few times we have had people suggest that there could be a bugs/reproducibility hazard! Including once at Mozfest.

Currently, I would say that the most closely aligned label is Danger of misuse. My thinking is that this includes things like "Used a statistical method without meeting the assumptions (and it matters)" or "Used to predict " but it doesn't work. HOWEVER, maybe this is too broad a hazard and we could stick to danger of misuse for situations where outputs are being used unethically, e.g. using a Deep Fake to frame them for a crime.

I think the image could be like a robotic ladybird or something.

Any thoughts?

[NatalieZelenka]

I think this could also fit in with Ismael's AI hype suggestion in that expectations do not equal reality.

[ninadicara]

Note to self, this may be combined/subsumed by Susana's paper suggestion of 'Potential for Faulty Result'

In terms of the ontology file this would fit nicely under General Data Hazard.