When we talk about the regulation of artificial intelligence, we often think of extensive legislation. The EU regulation on artificial intelligence, for example, is perhaps the first thing that comes to mind. However, it will take some time for its effects to become visible. The implementation process is gradual and will take several months. It will not be completed until two years after its adoption. However, there are other regulations that govern the AI market, even if they do not specifically mention the technology.
ISO standards are a good example of this type of regulation. They are widely recognized internationally, but are not part of the laws adopted by states or regional governments. There are thousands of standards, and some are of specific interest to the AI market. ISO/IEC 17021-1:2015 is one of them. It specifies the requirements that organizations must meet when conducting audits and assessments of management systems implemented by other organizations.
For AI, this standard will have an impact on the development and use of products and models. There are standards for products and services, but ISO/IEC 17021-1:2015 concerns processes in general. This means that companies developing an AI system will need to look to a company that complies with this standard if they want to be certified.
Only these companies that have the standard will be qualified to certify the processes of those developing or operating an AI model. In this way, ISO/IEC 17021-1:2015 will indirectly impose a certain level of quality on the AI market.
Call for excellence and uniformity
The ISO standard will act as a driving force for project quality in the AI market. By encouraging rigorous audits, it will stimulate the development of competent models. Not only because they are useful for users, but also because they will provide guarantees from a technical point of view.
This means that they will have to meet certain quality standards, which is crucial in areas such as security or data protection. In other words, it's about more than just performance, an area in which there will also be an incentive to achieve excellence.
It is likely that the facilities that hold the standard and certify other companies will look at certain factors when assessing them. One of these will be dedicated staff, so it is important to have a talented technical team behind an AI system. However, assessed companies must also have clear processes in place for the development and deployment of an AI service. This means that they must have defined guidelines for implementation, operation, risk control and problem resolution.
Considering the advancement in the AI market, another aspect that certification bodies can look out for in AI companies is their ability to train their staff. Continuous training of technical staff is key to ensuring that they are considered well prepared for the challenges of the technology.
All of this will in turn ensure that the products manufactured or operated by the company being assessed are of high quality. For this to be the case, however, there is another crucial factor: consistency. The standard states that all companies must be audited in the same way. Consequently, the guarantees offered by the systems of the certified companies - each in its own field - must be equivalent. And they will be, because they are measured by the same yardstick.
This is a way to minimize the bounce and improvisations of companies that are not prepared to develop or operate AI models. It will be positive for the market that there is a certain framework for collecting data and training systems.
A company that has been certified by another company that holds ISO/IEC 17021-1:2015 will be able to gain certain advantages in the market. Firstly, it will gain more trust from its customers. But this credibility will also be passed on to regulators and other bodies, such as vendors and potential partners. Ultimately, it will be driven by greater business opportunities.
Promoting ethical AI
One of the requirements that companies must meet with the ISO/IEC 17021-1:2015 standard is impartiality. This means that they must be independent and have no conflicts of interest with the companies they assess. And this independence is fundamental when it comes to giving the green light for an algorithm to work.
In AI, increasing attention is being paid to the potential biases of AI models. When it comes to people, these can lead to certain social groups being favored based on gender, ethnicity or other factors irrelevant to the purpose of the system. The principle of impartiality, in turn, means that all companies and their systems are measured by the same yardstick. As the standard encourages the pursuit of excellence in the schemes, it will also encourage the pursuit of excellence in all assessors.
Zertia
In a booming sector like artificial intelligence, where the benefits of its implementation are so high, the demand for companies capable of conducting audits that guarantee these types of ISO certifications is increasingly higher. Although this market niche is more developed in North America, pioneers like Zertia can be found in Spain.
This young auditing firm focuses on offering cutting-edge services to ensure that AI models and algorithms comply with the highest global standards and regulations. Holding ISO certifications not only demonstrates quality but also ensures greater business confidence, guaranteeing that the use of AI is fair, transparent, scalable, explainable, and sustainable.
