What can the IEC do to strengthen trust in new technologies?

DKE: The IEC actively takes the position of an international institution trusted worldwide due to various aspects such as diversity and the consensus principle. From your perspective, what is the key to winning such trust?

Schönhof: In my view, it is the main responsibility of any standardization organization to provide a platform for exchange to stakeholders from different sectors. On this basis, good, harmonized solutions to key issues can be found. It is the diversity that makes such solutions equally relevant to markets and societies.

Ayaz: I, too, see diversity as a key to trust as it is essential for achieving credibility and reliability. Diversity, as I understand it, has regional, cultural, industrial and regulatory aspects. The IEC, as a unifying element, has the responsibility to bring together all perspectives in order to develop standards that will serve the people, ensure safety and quality, and represent openness and technological advancement.

Hayes: I agree by adding a specific example: What the lack of diversity means becomes apparent in the design of crash test dummies and other safety equipment exclusively based on the average male from the western hemisphere. The body frames of females or people from other regions which may be smaller and more petite or bigger and more stout are not taken into account. The consequence is a lack of safety for large parts of the global population. That means that we, the IEC, need diversity in every aspect in order to develop functional and reliable standards.

DKE: AI is one of the main future technologies. In many areas, however, its misuse as well as AI hallucinations are a problem and can also undermine the trust in standardization. In its role as a standardization organization, the IEC should address this problem – but how?

Ayaz:

AI cannot be regulated, but we have to create the framework conditions for its use. Therefore, the focus is not on the algorithm, instead we are talking about traceability, data lineage, transparency requirements and robustness. We need mandatory labeling of AI-generated content and specified boundaries of use.

While generative AI operates by using every available source, industrial applications should use models working with a reduced amount of high-quality data. It must be clear, what AI is able and unable to do – and it must be part of its training that the purpose of the system is not to please the user but to deliver correct results. All this and more can be discussed and defined by standardization to provide clarity to manufacturers and users.

Schönhof: One of the key aspects, in my view, is how expectations are defined and handled. I think that one main aspect is the dealing with and definition of expectations. Let us assume a neural network intended to control a vehicle. Now, imagine a blue, one-eyed rabbit at the roadside – how is AI supposed to respond to such an unforeseeable situation? Standardization needs to transform our expectations into guidelines in order to create state-of-the-art AI.

AI hallucinations, in particular, are an inherent and therefore unavoidable part of the models. The IEC can merely specify guidelines on how to mitigate and manage these hallucinations. Misuse of technology, too, cannot be prevented. Even a screwdriver can cause damage if used outside the framework conditions of its intended purpose. However, standardization can provide basic principles for ways to develop AI that is resistant to misuse.

… and orientation for the handling of AI

Hayes: From my perspective, the IEC does exactly what Raoul and Batuhan are asking: With ISO/IEC JTC 1/SC 42 (Artificial Intelligence), experts provide a foundation and framework for the functioning of AI as a technology. This involves the development of standards on AI and of conformity assessment systems for the testing and evaluation of AI. 

A different consideration, in my view, is how we as humans use artificial intelligence. When trusting it blindly and without scrutiny, we will not obtain any valid results. The IEC has already specified corresponding guidelines which only have to be made available to the public.

IEC General Meeting 2026

The DKE is honored to host the annual event for international electrotechnical standardization. Under the title “Global Development. Driven by Standards.”, around 3,500 guests are expected in Hamburg in November 2026.

DKE: Let us take a look at the black box problem of AI and the fact that its decision-making channels are not always transparent to humans.  How can we develop standards such as to create a future basis for controlling artificial intelligence and thereby building confidence? 

Schönhof: In my view, the large models are inexplicable. Nevertheless, we can analyze how a certain result is obtained and we can classify and structure such results. This enables the interpretation of AI.

Standardization can also define quality requirements for the results of AI applications that can be implemented by the industry and used for conformity testing. This can build trust which will be passed on to the customer, namely the user – because we establish a quality infrastructure for AI. Some voices claim that standardization is too slow for this technology. There may be some truth in that, but as I view it: Without standardization, it will go wrong.

Ayaz: Besides the question regarding the explicability of AI, I do see its functional principle as a trade secret of the manufacturer. Nevertheless, the European AI Act requires the disclosure of the model for highly complex systems. Otherwise, I agree with Raoul – we should focus on processes, data, and results if we are to establish framework conditions for AI applications. This will not make AI entirely explicable but keep it controllable.

… instead of aiming for absolute control

Hayes: The wish for absolute control reminds me of the beginnings of cybersecurity. When software platforms and networked structures first kicked off, doubts and fears were running high. What we learned then is that, while complete control is impossible, we can establish suitable framework conditions to ensure the functioning of systems and components.

As already outlined by Raoul and Batuhan, our approach to AI should be based on the respective application. On this basis, we can also decide on the required extent of the framework. When AI is used by streaming platforms for content recommendations to users, this situation is less security-relevant than that of AI-supported monitoring cycles for components in the production of electronic equipment. The more severe the consequences of failures, the more guiding principles are required for the individual AI application.

DKE: What is the role of global standardization efforts regarding controllability of future technologies?

Hayes: Standards are the foundation for the successful development of new technologies. In order to provide efficient guidance, we need to prioritize and to establish principles for choosing the point at which a technology is ready for standardization, for example by means of the Standard Readiness Framework. Additionally, standardization must create a balance between safety and flexibility providing businesses with boundaries that are neither too wide nor too confined. 

Schönhof: We need research organizations worldwide which will delegate experts to committees on exactly those subjects in which the industry is not yet represented. Globally, we are dealing with a large bunch of technologies rolled out for billions of people. For that reason alone, I consider standardization immensely important – whether with respect to 6G mobile communications, energy storage, hydrogen or to solid state batteries. The sooner trust is won by standards, the easier will be the path beyond. 

Ayaz: Standards provide the industry with investment security, interoperability and acceptance among users, employees and, simply, society so that systems will actually be applied. Without standards, we will have technological proliferation, be unable of scaling and return to specific solutions for each individual problem. Those aspects are just as applicable to the rapidly developing new technologies as to any others.

DKE: What can the IEC do in general to improve its position as a solution provider to inspire trust and maintain reliability?

Schönhof: I think that there is confidence in standardization. Without question. Standards are implemented and thus confidence extends up to the end user – we must proceed on this path and not be deterred. However, many do not understand how standards are developed. The doors are open but only few know how to enter. In my view confidence can be maintained and enhanced by increasing active transparency.

Ayaz: We are experiencing a tense market situation. In the industrial sector, particularly German and European businesses are withdrawing from standardization. What is required to change this? We need to involve major corporations as well as medium-sized businesses at an early stage by lowering thresholds and by inviting and encouraging exchange. We must use pilot projects to make visible what we are doing and why. For this purpose, we can use AI in order to reduce complexity and to involve instead of outdistance others. This basis work is the essence of today’s and tomorrow’s standardization and the prerequisite for trust.

Hayes: This is an important point – I, too, think that we must continue to focus on a sound, technical, and science-based standardization where all voices are heard. Sometimes, the stakeholders at the table may not be the right ones. In this case, it is reasonable to wait until the right experts have been brought in order to hear all relevant aspects and remain reliable.

The IEC is the international organization that develops standards for the global electronics industry which drastically simplify global trade. We must develop our organization further and adapt our processes and tools to the requirements in order to maintain trust. Because once lost, it will be very difficult to regain.

Editorial note:

The responses reflect the interviewee’s personal views and opinions and do not necessarily reflect those of the DKE.

The original English-language article appeared on https://www.gm2026.iec.ch/

Basics of standardization: explained briefly and clearly

To understand the principles and fundamentals of standardization, you only need to know a few things. The basics include definitions, basic terms and explanations of processes and objectives, as well as further information on norms and standards.