In the technologically advanced world we live in today, data is incredibly precious. However, data is frequently underutilized or left unused. Though the causes differ, the issue is by no means new. An essay that was first published in the Harvard Business Review emphasizes how difficult it is for firms to manage and analyze the data that is now available.
In industry, where underutilized data can have a bigger cost impact, this issue is particularly noticeable. A World Economic Forum paper claims that efficient data utilization in manufacturing contributes to the sustainability and profitability of companies.
Oleg Savin, Unilever’s MES Expert on “AI in Business,” elaborated on the issue of underutilized data in production settings.
Two main conclusions from the second half of their chat will be the subject of the essay that follows:
Recognizing the barriers to the use of AI in manufacturing enhancing decision-making through the efficient use of a larger proportion of gathered data through the use of structured ontologies.
Aligning theory and practice to improve AI-driven manufacturing: utilizing a strong theoretical framework that combines operational and informational technology to successfully execute AI projects.
Quick Acknowledgment:
Recognizing The Barriers To Manufacturing AI Adoption
Savin outlines the areas in the manufacturing sector where data is underappreciated. He points out that although manufacturing systems do gather petabytes of data, only a small portion of that data—typically no more than 8–10%—is utilized to support decision-making or extract knowledge.
The main difficulty lies in the ability to separate relevant information from large, probably unstructured databases. Systems that employ ontologies, or organized models of knowledge, to extract pertinent data, identify trends, and assist in decision-making are at the heart of the future of artificial intelligence, according to Savin. “It will extract the necessary data and produce the patterns needed for making decisions,” Savin says, citing the ontological model.
According to Savin, companies should view data as a vital resource. Although intelligent systems must be able to retrieve pertinent data based on contextual understanding, there are obstacles in the way. According to Savin, there are two primary obstacles preventing the use of AI in manufacturing:
Misconception about AI: Neural networks are only one aspect of artificial intelligence, despite the all too often equating of the two.
Absence of platforms: Research and development is still ongoing on unified frameworks that directly design and build intelligent systems. Widespread adoption is restricted in the absence of the necessary instruments. Platforms that can automate numerous stages of development are extremely beneficial to manufacturing staff, including engineers, managers, and shop floor workers.
These frameworks have the ability to automate a variety of business and manufacturing tasks, including requirement formulation, domain analysis, requirement formulation and description, programming, installation, and implementation.
In response to a question about how company executives may better comprehend and use these technologies, Savin highlighted four crucial areas that executives must be knowledgeable about. These consist of:
Philosophy: Contains logic, epistemology, and phenomenology.
Semantics is a component of linguistics.
Cognitive science explains how artificial intelligence should learn.
Systems theory, cybernetics, and fractal models are all included in mathematics, which is the language of description.
He discusses how the aforementioned domains are integrated in the fractal cognitive computational model of consciousness, which artificial intelligence ought to emulate. The model is fractal in essence because it mimics the organizational patterns found in biological awareness.
Savin provides direction. When asked what inquiries a manufacturing executive in charge of a startup AI project ought to make. These inquiries consist of:
Which essential measures, or core indications, best describe the stability or success of current processes?
Can the AI system demonstrate that it can forecast deviations and maintain stability?
Are the choices made by the system measurable and explicable?
Can ontologies be used to change the system instead of only retraining it on fresh data?
Savin continues by outlining the three main integrated indicator dimensions that make up the fractal technique. These consist of:
- Time
- Value
- Efficiency
Enhancing AI-Powered Manufacturing Through Theory And Practice Alignment
Process optimization, quality control, and predictive maintenance are all aided by AI applications in manufacturing processes. Implementing AI will inevitably present certain difficulties, such as the expense of classifying massive data sets, but Savin tackles a more fundamental problem.
Savin’s statement, “There is no practice without good theory,” encapsulates the significance of coordinating theory and practice. He highlights the importance of having a strong theoretical base and lists the four advantages that follow:
IT expertise will become less important as management turns its attention to line employees’ motivation. Savin believes that motivation ought to foster and encourage creativity.
Information technology (IT) and operational technology (OT) convergence: When attempting to model or configure systems across many settings, including in production or the cloud, current IT design techniques have limitations. Operations are now more affordable, effective, and convenient as a result of the convergence of IT and OT.
Self-organization and self-management, as well as end-to-end value-creation processes, will be made possible by smart operating systems, which will enable organizations to transform data into useful resources.
Reaching the process stability level needed for human time horizons: Accounting, control, and management are examples of traditional functions that can become obsolete.