AI Norms and Standardisation – the road ahead

In a world were products and services are traded with light speed, trust between partners is a key ingredient. Society, institutional and economical partners has developed norms and standards willing to promote and ensure compatibility and agreed processes. Those activities include terms and their semantics, validation conditions, quality gates, measurements, safety conditions, to mention just a few of them.

With respect to Artificial Intelligence, we could observe that its role along various human related activities exposes some special features. First and obvious one is related to the nature of AI against the governance process which is expected to be executed with ethics and privacy conditions and enable the trust between involved actors.

The AI landscape is complex, and standards and regulations need to carefully structure actions and measurements in order to provide a coherent and appliable framework aiming for trustworthiness.

The European plan for AI regulation took a careful stance and is now represented by the EU Artificial Intelligence Act, the first proposal of this regulation was published in April 2021 (https://artificialintelligenceact.eu).

In simple terms the policy objectives express the objectives to be followed:

• Ensure that AI Systems on EU market are safe and respect EU laws and values
• Create legal certainty to facilitate investment and innovation in AI
• Enhance the governance and enforcement of existing legal requirements
• Facilitate the safe, lawful and ethical development of AI applications

This regulation sets the space for few key relevant aspects:

• Harmonized rules for the placing on the market, putting on the service and the use of AI systems in EU
• Prohibitions of certain AI practices
• Specific requirements for high-risk AI systems and obligations for operators of such systems
• Harmonized transparency rules for AI systems intended to interact with natural persons, emotion recognition systems and biometric categorization systems, and AI systems used to generate or manipulate image, audio or video content.
• Rules on market monitoring and surveillance

The objectives of this regulation are: to ensure that AI systems on the EU market are safe and respect EU laws and values, create legal certainty to facilitate investment and innovation in AI and l facilitate the safe, lawful and ethical development of AI applications in the EU. In its current form, the regulation aims to encourage technical regulations such as allocation of responsibilities in the AI value chain between providers, users, brokers, etc.

Yet where is STAR located? The project comes in a sweet spot due to a number of arguments.  STAR, by design, considers that AI systems on the EU market are safe and respect the EU laws and values, create legal certainty to facilitate investment and lawful and ethical development of AI applications.

Industrial use of AI, especially under the Industry 4.0 umbrella, aim to provide innovation and productivity increase on areas where humans and machines collaborate such as advanced manufacturing. STAR project use cases and architecture components demonstrate that use of AI can be safe and effective if the AI solutions consider the observability and validation of shopfloor planning and actions.

Likely the EU AI Act should consider industrial and B2B applications and distinct it from B2C applications and have a close attention to industrial/machine data. This aspect is confirmed by the interest declared by the European Commission and the mission given to the European Committee for Standardization (CEN) and the European Committee for Electrotechnical Standardization (CENELEC) to scrutinise and drive European standardisation on this aspect (see here). At the same time we should be aware that the AI Act’s declared scope is consistently vaster than industrial landscape on its current form. That means that we can only consider use of AI as a high-risk factor when health, safety and security are under scrutiny.

The STAR project admits that a “continuous compliance” with current regulations in place approach is valid and useful. The project’s efforts do not match just a technological promise, but also a plan for evolutionary move towards safe and reliable innovation in manufacturing.

By: Septimiu Nechifor, SIEMENS SRL