Bernhard R. Martin answers questions of ethics in the application of artificial intelligence and current developments within the EU in a new working paper from the AF Institute. AI systems have the potential to contribute to the management of major problems such as climate change and environmental degradation, stresses the Vienna based media-sociologist.
Responsible handling of artificial intelligence is important, for example to avoid discrimination on the basis of ethical affiliation. The ethical and legal assessment is complex, however, because AI systems can develop themselves due to machine learning and supply chains can be complex.
There is a risk of fragmentation of the internal market by national regulations regarding AI. A legal framework is important, but the flood of regulations would affect SMEs in particular. Regulation could already begin with the data used for training, for example by using representative data sets (gender, ethnicity).
Letting people decide
It is particularly important not to leave decisions to machines. In the AF institute’s working paper, Martin outlines four options for human supervision:
- Results delivered by an AI system only become effective if they have been reviewed and validated by a human
- Results delivered by an AI system become effective immediately, but human intervention is ensured at a later stage
- Real-time monitoring of AI systems during operation
- Limitations on the AI system in the design phase
Artificial intelligence increases productivity
“AI offers important efficiency and productivity gains that can strengthen the competitiveness of European industry and improve the well-being of its citizens,” Martin concludes.
Access to the essay by Bernhard R. Martin is available in the download area. Registration required.