AI is a technology that is currently in many heads around the world. Border are transparent for technology and are becoming transparent for data. On the other hand AI can and probably will affect everyone. Sensible and clear regulations are needed to balance the benefits and threads of AI.
We focus here on AI in general. Specific applications as killer drones shall be regulated specifically, e.g. via the CCW - LAWS conventions. Also data protection shall be regulated separately - but in coordination with AI.
The Climate convention (Paris convention) is an example that took and takes long to reach global relevance and acceptance. As the existance of AI is not debated, the adoption of a regulation on AI should be faster. But this is also needed because the evolution of AI is markedly faster.
The challenges of AI are observable and immediate for many in todays' life. AI also attracts more economic - business and national - interests which could tip the adoption of the regulation due to the influence of this large national or international players.
The regulation of AI should be international and it should foresee/predict/??? future developments of AI.
As AI may influence the individual's freedom, the regulation should ensure this freedom. Establishing the regulation of AI in the sense of a universal human right - along with the established and accepted ones on right to fair process, no penalty without a law, innocent until proven guilty. Limitation of the power that AI may have on the individual humans also affects the human dignity, which is a universal / global value.
The speed and high technicality of AI will require the involvement of experts early and continuously.
The exact implementation of the AI regulation has to be discussed and is not yet solved. The key challenge here is the balance between the adoption and the possibility for enforcement.
- Adding the right to an established convention might make enforcement easier.
- Creating a new convention would give it more flexibility and - given a good and modern organization - the possibility to adapt quickly to the evolution of AI.
- A more bottom-up approach would be the (voluntary) commitment of AI users (e.g. goverments, companies) to a globally accepted code of conduct.
- Enforcement in this case would rely on the citizen's reaction to breach of the commitment.
- This approach might allow for a faster adaptation from the experts in the field.
- Another proposed idea (from Erdelyi and Goldsmith 2018) is to form an inter-governmental organization.