SNCF Innovates with AI for Enhanced Railway Maintenance and Document Management

Embracing the Digital Revolution: With a massive workforce of 250,000, SNCF operates 15,000 trains daily, catering to 5 million passengers. Their digital infrastructure is impressive, boasting a 20,000-kilometer fiber optic network across all stations and sites. The high-capacity telecom infrastructure, capable of 20 to 400 Gbps speeds, handles terabytes of data from IoT sensors every day. Christophe Fanichet, the group’s digital deputy CEO and CEO of SNCF Voyageurs, highlights that this robust setup ushered the company into the AI era.

Adapting to Generative AI: After observing ChatGPT’s emergence, SNCF initially allowed unrestricted access to the tool for its employees. By late 2022, a daily average of 500 to 2,000 employees was using the service. Considering data protection, especially for sensitive information like transport plans and maintenance documents, SNCF swiftly created their private version of ChatGPT, dubbed SNCF Group GPT, hosted on Azure OpenAI. Exclusive to executives and key managers, the company also established a training program to optimize usage.

Customized ChatGPT Support for Staff: The platform includes a document space with a Reference Application Guide (RAG) that amalgamates over 70,000 documents to assist ground workers, for example, by simplifying access to technical documentation during maintenance. Another feature allows employees to upload documents, aiding white-collar workers in synthesizing and delegating tasks across various departments.

As of now, over 2,500 employees use SNCF Group GPT daily, with expectations to reach 10,000 by year’s end. Additionally, SNCF is developing a translation tool, Trad SNCF, spanning 130 languages and planned to be available to over 50,000 workers.

Generative AI in Industrial Projects: In collaboration with French startup Mistral, SNCF’s industrial projects include enhancing passenger information during disruptions. The predictive technology provided by Mistrial’s Large Language Model (LLM) leverages historical data to improve communication with travelers, refining estimates of recovery time by considering various factors.

Future Developments and AI’s Role in Decision Making: The SNCF also explores AI’s potential in transport planning optimization and maintenance scheduling, where AI will support solution designs and alternatives. Meanwhile, they utilize generative AI to analyze railway infrastructure photos, spotting potential train defects. Humans, however, retain decision-making authority over machine-identified issues. Lastly, SNCF is exploring a voice bot for their information hotline, aiming at a seamless conversational experience.

Important Questions and Answers about SNCF’s AI Initiatives:
1. What is the role of AI in SNCF’s operations?
AI at SNCF is being integrated for a variety of applications, including maintenance optimization, document management, and enhanced passenger communication during service disruptions. It acts as a support system for decision-making processes and facilitates the daily workflow of employees.

2. How does SNCF ensure data protection when using AI technologies?
After initially offering unrestricted access to ChatGPT, SNCF recognized the need for data protection and created a private version hosted on Azure OpenAI. This version, SNCF Group GPT, is exclusive to executives and key managers, safeguarding sensitive information and compliance with data protection regulations.

3. What are the potential future developments for AI at SNCF?
SNCF intends to further exploit AI in areas such as transport planning optimization and maintenance scheduling. They are also considering the implementation of a voice bot for customer information services to enhance the conversational experience.

Key Challenges and Controversies:
Data Security: The integration of AI requires handling large volumes of sensitive data, and the company must continuously ensure that this data is protected against cyber threats.

Job Impacts: The introduction of AI can fuel worries about job displacement or changes in workforce requirements, necessitating careful management and training programs.

Trust and Reliability: Relying on AI for critical infrastructure such as rail networks raises concerns about the trustworthiness and reliability of AI-driven decisions.

Operational Efficiency: AI can analyze vast amounts of data, forecast maintenance needs, and optimize operational tasks, leading to increased efficiency.
Ease of Access: With AI-powered tools, SNCF employees can more easily navigate and utilize the vast collection of documents necessary for their work.
Enhanced Communication: AI improves passenger information systems, especially during disruptions, which can lead to a better customer experience.

Complex Implementation: Integrating AI into an extensive and vital infrastructure like a railway system is complex and requires significant resources.
Initial Costs: The initial development and implementation of AI tools can be costly, although these expenses may be offset by long-term savings.
Overreliance Risk: There is a potential risk of becoming overreliant on AI systems, which could lead to vulnerabilities if they fail.

For further information on SNCF, you can visit their official website at SNCF. Please note that accessing the site provides the most current and relevant information directly from the source.

Privacy policy