Odgovornost države za neuporabo oziroma neustrezno uporabo umetne inteligence pri preprečevanju poplav in vplivi na človekove pravice

slovenian

2023

volume 49, issue 6/7

Lexpera d. o. o.
(Obvezni izvod spletne publikacije)

Scientists warn that river flooding caused by climate change is one of the most prevalent natural disasters, which is due to increase in frequency. Flooding is by some accounts the most devastating natural disaster occurring worldwide, resulting in the destruction of infrastructure and farmland and has serious impacts on human health and life. Unfortunately, floods cannot be prevented, but they can be anticipated and prepared for. As part of adapting to future floods, several artificial intelligence (AI) systems are being developed that analyse data ranging from rainfall, river flow, geological structure of the river basin, satellite imagery and past flood events to produce simulations of future flood models. These can be used to predict several days in advance how river levels will rise in the event of more severe rainfall, and the extent and severity of flooding. Their purpose is to help national decision makers to take appropriate measures to prevent or at least reduce flood damage and to keep the affected population properly informed. This paper presents different forms of AI systems for flood forecasting and prevention. It highlights a number of challenges in the use of these systems, such as bias, lack of data, inexplicability of the obtained results and their unpredictability. These challenges affect the obtained predictions and also affect the (lack of) confidence of decision makers in these obtained outcomes. This may culminate in decision makers' failure to take into account the obtained predictions from AI systems and their failure to adopt adequate measure to respond to future floods. If this happens, who is responsible for the ensued damage to property, the impact on human health and life, and the violation of fundamental human rights?

09.05.2024