Concern in the UK over the growing use of artificial intelligence in public administration

An investigation by the British newspaper The Guardian reveals that eight ministries delegate some of their tasks to artificial intelligence.

In some cases, they do so for mainly practical reasons and do not have a significant impact, such as for automatic passport control at airports.

The Ministry of Employment, however, uses algorithms to allocate aid and allowances, filtering out abusive requests. Police facial recognition software is also automatic and the Home Office uses the technology to validate marriage licenses and identify fake marriages.

Some of these artificial intelligences are sophisticated, not always perfectly understood by the employees of these services and, as The Guardian points out, not always well managed.

According to independent legal experts, the United Kingdom is heading towards a state scandal because of these systems which tend to reproduce social discrimination.

Artificial intelligence systems “learn” from extremely large data sets, drawn from cases previously reviewed by humans. One of the consequences of this type of learning is the mimicry and amplification of prejudices already existing in society.

In the case of benefits, for example, one MP explained that more than a dozen people in her constituency had had their payments suspended for no reason. Their common point was that they were almost all of Bulgarian nationality.

When it comes to police facial recognition, black faces stand out as “suspicious” more often than white faces.

Several law and ethics specialists denounce the fact that these automatisms are not held responsible. Last month, the government dismantled the Center for Data Ethics and Innovation, an independent consultancy group specializing in AI.

However, artificial intelligence is one of the current government’s favorite subjects. Prime Minister Rishi Sunak himself is promoting the use of AI in public services. This could improve the accuracy and speed of medical examinations, or save teachers hours of work preparing their lessons.

Experts recognize that there is huge potential, but it comes with risks. On November 1-2, Sunak is hosting an international summit on “AI and Security” in north London, with industry players from around the world. Associations are calling for more transparency and regulation in this area.