Olaf Scholz visited the Federal Office for Migration and Refugees in Nuremberg. The chancellor said he could imagine the use of AI in the processing of asylum applications. This is also being worked on in several places. According to dpa, Scholz spoke of “regular decisions” that can be made “quickly and yet with good quality”. The speed and quality of asylum decisions are at the heart of his approval.
Advertisement
In principle, the authority should be given more money for digitalisation. This includes the development of AI applications. It is important that initial reception centres in the federal states start asylum procedures. This often happens only after refugees have been distributed among the municipalities. Administrative courts also need to be further accelerated. The president of the Federal Office, Hans-Eckhard Sommer, also said at the meeting that the authority’s workload was high – although the situation for asylum procedures was more comfortable this year than in previous years. The average processing time for initial asylum applications is 4.6 months.
AI deemed highly risky in the asylum process
However, using artificial intelligence in asylum procedures may not be so easy. The AI Act provides for a particularly large number of regulations here, as asylum procedures are considered applications with a high level of risk. People seeking asylum are in a precarious situation and are dependent on decisions. This is why the accuracy and transparency of these systems is particularly important. It must be ensured that AI systems protect the fundamental rights of those affected – such as the protection of private life and personal data, as stated in the law.
This means that the use of AI for border control or later for processing asylum applications is not ruled out, but it must meet certain conditions. For example, the problems arising from the reproduction of stereotypes and prejudice in AI systems must be considered. High-risk systems must be particularly transparent, processes must be accurately documented, there must be risk management and, ultimately, human supervision.
Read this also
(EMW)
