How large language models can boost industrial automation
Published: 14 March, 2024
ChatGPT demonstrates the potential impact that large language models could have on the way people interact with machines. According to Boston Consulting Group, generative AI will surely bring about a boost in productivity from people and machines working together; and there is no place where people and machines interact more than in industrial automation segments. Aurelien Le Sant, Chief Technology Officer, Industrial Automation, Schneider Electric explains further
Large language models are created using deep learning algorithms that are trained in massive amounts of text data. Recently, large language models have spread into various industries and increasingly applied in industrial automation. The ability of these models to understand and generate human language can greatly improve the efficiency of industrial processes. Our own conservative estimations show the potential of large language models to eliminate 20% of the effort required by an OEM to build a machine PLC application.
The move from GPT-2 to GPT-4 enables models to handle more content as inputs, exponentially increasing the applications of large language models in industrial automation. Customising models for specific use cases or tasks within industrial automation can include:
1. Code generation, documentation, refactoring, and testing
2. Natural language interfaces
3. Automation system design and development However, it is important to consider the limitations that may be prevalent before the large language models are industrialised.
1. Code generation, documentation, refactoring, and testing
Large language models can be used to generate code for control systems such as PLCs or to generate HMI screen using natural language inputs. This reduces the time and effort required to develop control applications. Furthermore, large language models have the potential to improve the quality of generated code, leading to fewer errors and faster commissioning times.
Another application can be the automatic generation of recipe code which would save time when changing parameters, suppliers, or ingredients. The time to create the recipe often impacts production time so any saving that could made here would increase efficiency. Large language models can also automatically generate documentation associated with the code – like automatic test scripts which has always been time-consuming for the operator.
Schneider Electric has been testing the use of private access versions of large language models to train on our own EcoStruxure Machine Expert applications. The results are promising, with code able to be generated quickly and somewhat accurately, still needing a human eye to review, but we can see that the same kind of model can be applied to other software applications.
2. Natural language interfaces
Large language models can also be used to create natural language interfaces for industrial automation systems, allowing operators to interact with these systems being human language rather than programming languages. This capability enables operators to rapidly access existing documentation using natural language commands. As documentation is digitalised and fed into secure and specific large language models, operators can simply ask questions. For example, “what does error code 8975 mean and how can I resolve it?”, the model will draw from approved and official manuals, technical descriptions, and source codes, for quick answers.
An often-reported challenge across industrial businesses has been that domain knowledge can reside with specific people. While this is being addressed through digital transformation of work practices, many factories still encounter maintenance issues that can only be fixed by one person. If that same operator’s expert knowledge can be fed into the safe and secure language models, operators could use voice commands to troubleshoot and take corrective actions. The same large language models can also be applied to the creation of user manuals and documentation for machines, automation products, and systems, freeing up time for engineers to apply their knowledge to value-adding activities.
3. Automation system design and development
Whether it is an expansion or a greenfield development, the design of automation systems requires significant coordination between numerous vendor and customer departments, and often third parties like regulatory authorities. Large language models can standardise known parameters to reduce the time taken and add a competitive advantage. Customer and partner inputs can take months, or even years, of rigorous work from experts to ensure quality and viability. Large language models speed up the whole process while adhering to strict compliance at all levels.
Ethical considerations and limitations
The use of large language models in industrial automation raises several ethical considerations and risks which must be factored in to ensure that the technology is used in a responsible manner.
Safety: If AI models are used to perform actions in industrial automation, clear safety practices must be the first consideration.
Data privacy: Large language models require large amounts of data to be trained. This data can include sensitive information that must be protected, adhering to all GDPR compliances.
Bias: Large language models can amplify societal biases in data used for training. This could lead to unfair and discriminatory outcomes, it’s crucial to identify and mitigate these to ensure fair and equitable outcomes.
Security: Large language modes can be vulnerable to malicious attacks such as model stealing or adversarial attacks. Models must be protected against these threats.
Explanation: Large language models can be difficult to interpret, sometimes making it a challenge to explain responses. This can be problematic when used in decision-making processes as all decisions must be understood, ensuring they are fair and reasonable.
People focus: Even with the capabilities of large language models, human interaction is still essential. These models must complement human capabilities, with any response being checked by experts before it is used.
Industrialisation of large language models
To fully leverage the benefits of large language models, they must be deployed appropriately while considering all the ethical considerations and limitations. Another important aspect to consider is the potential base size of that large language model. With each development, for an effective execution of large language models would require an enormous amount of compute power and space.
The use of large language models is becoming increasingly common across a wide range of industries. McKinsey notes that “generative AI has the potential to change the anatomy of work, augmenting the capabilities of individual workers by automating some of their individual activities.”
Large language models like GPT-4 are changing work practices in industrial automation by using them for code generation, with natural language interfaces and, further into the future, for automation system design and development. Using advanced machine learning techniques to generate high-quality code and documentation quickly significantly increases efficiency and reduces errors, when mindful of the ethical and risk considerations. Industrial automation can harness the potential power of large language models to change the way work is done across the complete lifecycle from design and build to operate and maintain.
For further information please visit: https://www.se.com/ww/en/
https://twitter.com/SchneiderElec
https://www.linkedin.com/company/schneider-electric/
![](/images//5769.photo.4.png)