AI mistakenly deprives people of social benefits

British officials are using artificial intelligence (AI) to make official decisions, according to an investigation. These include: the granting of social benefits and the issuing of marriage licenses.

Since the further development of artificial intelligence, many ethical questions have arisen. Does an AI have rights? Does it treat everyone equally – regardless of their background or attitude? And what are the dangers of such systems? It will probably take some time before there are answers to these questions.

A new case from Great Britain recently showed the power algorithms already have on people’s everyday lives. Because according to a report in the British newspaper The Guardian The government there uses artificial intelligence to make certain decisions. These include: the granting of social benefits and the issuing of marriage licenses.

Social benefits: British officials use AI for decision-making

The research shows that officers in at least eight departments and some police forces use AI systems to make decisions about immigration or legal issues. But the effects are even more far-reaching. Because artificial intelligence also seems to have problems with equal treatment.

This includes an algorithm for approving or withdrawing certain government benefits or a facial recognition tool. While the algorithm inadvertently deprived many people of their benefits, the camera system disadvantaged people with dark skin. The problems usually lie not in the algorithm, but in the training data on which the system is based.

Training data is not transparent or viewable

If there are certain prejudices or images about certain groups of people, the system reflects these in the corresponding decisions. The Guardian found, for example, that an immigration authority tool excessively discriminated against people from Albania, Greece, Romania and Bulgaria. However, those responsible assured that the systems were safe and fair.

But this highlights another problem. Because the algorithms are not transparent and were not allowed to be used Guardian can be viewed. Nowadays, AI development mostly takes place behind closed doors. People’s trust in computer systems is therefore often not particularly high.

Also interesting: