It has been made clear that the rules have to be established firmly by the EU and international institutions where a regular check for compliance is made by an external institution.
One of the rules that are very necessary is for example, whenever an AI interacts with a human being, identification is where the human colleague can have a longer level, for example, emergency stop switches. In the case of AI, it is decision-making powers where the human being must be given a right to ask natural people such as the works council or another responsible body in the company to review its AI decisions.
Operators of AI in addition to working with machine learning must be obliged to independently monitor the rewriting of algorithms where tests are run through. In the case of an AI-based robot, algorithms are checked regularly by the companies for the reason of operational safety alone. Also, it has to be kept in mind that AI algorithms are written by people who are regularly deepening their skills such as specialists through further training.
As with the case of the steam engine, devastating accidents can happen but a regular inspection by the forerunner by TüV Rheinland, the former Dampfkesselüberwachungsverein DÜV, led a large-scale technical plant, vehicles, and lifts that is becoming a safe device.
If the AI work needs to be concluded then without independent third-party monitoring and a lifelong further qualification of human specialists, it is going to be hard for using Artificial Intelligence.