New Whitepaper on AI Standards for Engineering
Data is literally the “new gold” of our time. While two zettabytes (equivalent to one billion terabytes) of data were generated worldwide in 2010, this figure had already risen to 47 zettabytes by 2020. Moreover, this immense growth is forecast to exceed 2142 zettabytes (in 2035) in the coming years. The use of artificial intelligence holds enormous potential for tapping into and utilizing this flood of data, especially for product creation. The white paper explores the question of which data standards can be used by common AI frameworks.
It shows that in addition to the amount of data, the number of end or producer devices is also increasing: According to Statista, the number of internet-connected products is estimated to reach 75 billion by 2025.
To tap and use this flood of data, the use of artificial intelligence holds enormous potential, especially for product creation. Successful use of AI can cut production costs, reduce development time and make optimal use of resources. For example, the use of AI is expected to increase profitability by an average of 38% by 2035.
In order to be able to fully develop the potential of artificial intelligence for product creation, the data from product creation plays a central role: The heterogeneous, complex data from the various software frameworks of engineering IT must be made usable for AI applications. Manufacturers of software solutions, especially in the AI field, are adapting to the increasing demand from industry. However, industry standards and those in the even newer “AI sector” differ, which is why collaboration between the two domains has not yet reached its full potential. We are therefore investigating the extent to which data standards commonly used in engineering IT can be read and used directly by the most common AI frameworks, and in which cases this may not work smoothly.
For this purpose, we have defined the term “AI Readiness” of a standard in the project, as the ability of the standard to be read in by defined AI frameworks or to be transformed into formats that have this ability. With this, we aim to provide a recommendation for the use of appropriate data standards from engineering IT for use in AI applications, as well as address specific challenges. This white paper provides an initial overview of the results of the analysis. The focus is on neutral, open standards to ensure a cross-tool and cross-vendor view.