Representative End User Clients
Representative Automation Clients
Representative Software Clients
The summit had four tracks that covered different aspects of AI evolution, such as future trends, trust and regulation, data challenges, and scaling up.
I also saw examples of how Industrial AI is transforming businesses, such as BMW Group's Cloud Data Hub and Kellanova's Digitally Connected Enterprise. One of the interesting topics was why companies are reluctant to share data and how data sharing can enable more innovation and value creation. On the second day, Siemens' CEO of Factory Automation explained Siemens vision for Industrial AI and how they use an Industrial Copilot to achieve maximum value.
One of the main topics of the summit was the definition and importance of industrial grade AI. Industrial grade AI is a term that Siemens uses to convey the message that AI needs to be reliable, robust, and scalable for industrial applications.
Unlike consumer or social media AI, where 80 percent or 90 percent accuracy might be acceptable, industrial AI requires a much higher degree of precision and quality, aim should be the five nines (99.9998 percent). You can't afford regular failures or erroneous outputs when you are dealing with critical processes, machines, or products. That's why Siemens has a high-quality ambition and sees it as its responsibility to help the ecosystem make AI industrial grade.
Siemens is not only a leader in delivering industrial grade AI solutions, but also in investing in AI talent and research. A recent study by Zeki ranked Siemens as the top European industry player in terms of hiring "AI staff with advanced skills", followed by Bosch. Siemens also has a strong research and innovation ecosystem that enables it to stay at the forefront of AI developments and trends. Siemens collaborates with leading academic and industrial partners and supports various initiatives and projects that aim to advance the state of the art in AI.
One of the emerging trends in AI research is the use of foundation models for many applications like robotics. Foundation models are large-scale models that can learn from a variety of data sources and tasks, and then be adapted or transferred to new domains or applications. Companies, like the robotic software company Intrinsic, presented practical examples of how foundation models are used for the development of robot skills, such as object pose estimation, robot grasp planning, and robot motion planning. These skills enable robots to perform complex and dynamic tasks, such as picking and placing objects, in a flexible and efficient manner, with less training than ever before.
Another highlight of the summit was the presentation of the Industrial Copilot, Siemens' vision for an advanced, AI-powered assistant tailored to the complexities of the shop floor, will be available soon for customers. The Industrial Copilot is designed to support operators, engineers, and managers in their daily tasks, such as monitoring, troubleshooting, optimizing, and planning. The Industrial Copilot can understand natural language, provide contextual information, and suggest actions based on data analysis and domain knowledge. The Industrial Copilot is not a replacement for human workers, but a partner that can augment their capabilities and enhance their productivity, security and safety.
The summit also addressed the impact of different regulations like the EU AI Act on industrial automation. The AI Act is a proposed regulation that introduces a risk management approach with different types of requirements to be met depending on the criticality of the AI application. The AI Act aims to ensure that AI is trustworthy, ethical, and human-centric, while also fostering innovation and competitiveness. The participants agreed and emphasized the need for a balanced and proportionate regulation that does not hamper the development and adoption of AI in the industrial sector.
Finally, the summit showcased how Siemens has in-place guidelines for deploying AI models in its network of factories while ensuring security and production performance. Siemens has developed a comprehensive AI governance model that covers the entire lifecycle of AI solutions, from development and deployment to operation and maintenance. The AI governance model includes aspects such as data quality, model validation, explainability, robustness, security, and monitoring. Siemens also leverages its own platforms and tools, such as the Industrial Edge and the AI Library, to facilitate the integration and management of AI models in the factory. Siemens has deployed more than 300 AI models in its factories, and has achieved significant improvements in quality, efficiency, and sustainability.
In short, some of the key takeaways for me were:
AI should be industrial-grade and easy to use
Federated learning can help transfer best practices across factories
Sparse data can be handled with synthetic data and reinforcement learning
The future is data sharing to increase efficiency and industrial grade results
Pre-trained models can make AI accessible for non-experts
Multimodal LLMs will gain importance in the future of industrial applications
People-centered AI involves a human in the loop to validate and release models
AI system health monitoring is essential to reduce risks
Data quality is crucial for AI value creation
The Siemens AI Summit 2024 was a remarkable event that demonstrated the value and potential of industrial grade AI for industry and society. Siemens is committed to making AI work for its customers and partners, and to creating a sustainable and intelligent future. I look forward to the Siemens AI Summit in 2025.