As we previously announced, our Foundation participates in the works of the Artificial Intelligence Working Group (GRAI) at the Chancellery of the Prime Minister (KPRM). We are working within the AI supervision subgroup and are trying to develop the best mechanism for quick verification of systems introduced to the market.
Based on the EU's Artificial Intelligence Act, seven administrative units can be distinguished:
- notifying authority,
- conformity assessment body,
- notified unit,
- market surveillance authority,
- law enforcement authority,
- the national supervisory authority and
- competent national authority.
Each of these entities is assigned different functions and tasks.
It is difficult to assign these functions to already existing administrative units due to the specialized nature of each sector. The multiplicity of authorities whose opinion would be necessary to accept an AI system would risk economic paralysis. The fragmentation of supervisory competences would hamper rapid technological development and would reduce competitive advantages.
The AI supervision subgroup is inclined to support appointing a new entity that would liaise between various public authorities - both national and European, and which would accept and supervise AI-based systems. The Foundation supports this idea of creating a new unit with appropriate competences to deal with this specialized and interdisciplinary area of AI systems. The appointment of a new body involves some financial and personal investment, but maintaining the highest possible level of neutrality of such a body speaks for its creation.