Following new regulations, data management has become more complex; however, the use of private artificial intelligence can help companies navigate this new legal landscape.
According to the latest AI Chamber report, 90% of companies in Central and Eastern Europe (CEE) are using artificial intelligence and machine learning[1]. Currently, both AI model providers and their contractors are considering ways to leverage new technological capabilities while maintaining compliance with regulations. Equinix, a global digital infrastructure firm, highlights that no organization is currently fully prepared to comply with the Artificial Intelligence Act, and company leaders have much work ahead to adapt their businesses to this new reality.
Lawmakers introduced new data management requirements under which data sets used for training algorithms should be predominantly representative, error-free, and complete [2]. Therefore, companies are obligated to more closely analyze the data used for training models.
In order for AI models to effectively operate, they must be trained on extensive data sets. The algorithms are therefore only as good as the data on which they are trained. The goal of the new regulations is to encourage organizations to more closely verify the origin of all new data. However, in times when companies constantly need to expand their data sets to train the best possible AI models, the regulations in their current form may hinder this. Using databases or AI models that are fully compliant with the new directives will be possible. On one hand, this will limit the possibilities for unrestricted experimentation and training of AI. On the other hand, it will safeguard against data with unacceptable or non-compliant parameters – says Sylwia Pyśkiewicz, Managing Director of Equinix in Poland.
For businesses, it is key that their fundamental AI models function smoothly whenever they are needed. Lawmakers have recognized this need, stating that high-risk AI systems will operate without disruption if contingency plans and backup systems are developed for them [3].
Companies can follow these guidelines by creating backups both locally, within their own servers, and in the cloud. Geographic diversification of data center access will play a significant role here, and collaboration with reliable digital infrastructure providers who offer global data center reach capable of supporting AI, as well as a well-developed ecosystem of cloud service providers, will become increasingly important.
The evolution of generative AI is prompting companies to change their thinking about data centers. Businesses need a well-optimized technical infrastructure that can handle an increasingly large amount of data without generating significant delays. Providers will need to demonstrate flexibility to connect AI platforms to the cloud as quickly as possible, which will allow companies to avoid downtime. We have market indicators suggesting that organizations are currently looking for data center providers that implement technologies maximizing server performance, such as liquid cooling or diversification of energy sources – adds Sylwia Pyśkiewicz.
For companies looking to capitalize on the potential of artificial intelligence, the trend toward increased regulatory complexity may be daunting. Many organizations perceive public cloud as the natural method for handling AI on their own since it does not require building in-house data centers. In contrast, some entities opt for hybrid actions, i.e., storing sensitive data in the private cloud while placing other information in the public cloud.
The widespread availability of data centers worldwide allows organizations to maintain flexibility and diversify the security of sensitive data. As a result, companies at every stage of AI implementation can avail exactly the solutions they require.
[1] AI not only for big corporations – 90% of SMEs use artificial intelligence (dlahandlu.pl)
[2] Article 10: Data and data management, EU Artificial Intelligence Act.
[3] Article 15: Accuracy, reliability, and cybersecurity, EU legal act on artificial intelligence.
Source: https://managerplus.pl/polskie-firmy-staja-przed-wyzwaniem-ai-act-co-musza-wiedziec-89467
