AI is the boost Mercedes-Benz needs to digitally transform

Mercedes-Benz has long relied on machine learning and classic AI. But now you’re also using generative AI, for example in the MO360 production environment. What exactly is it about and to what extent does it change the profiles of employees?

With digitization and the increasing use of powerful AI systems, job profiles are changing in production and administration. AI is intended to improve and facilitate the day-to-day work of employees. New digital tools, for example, further enable production employees to optimize processes and quality management in the long term. The first interim results in MO360 are very promising, especially since we see it’s not only accepted by the IT experts, but also by the masters on the shop floor.

Your company has started the Turn2Learn qualification initiative. Which AI or digitization skills are taught to the employees?

Turn2Learn is an initiative of our HR department that focuses on digitization and AI. The offer ranges from AI and machine learning for beginners, to the Prompt Engineering learning path, to training courses on programming languages ​​such as Python, deep learning and neural networks, reinforcement learning, RPA, and natural language processing. In total, employees have access to more than 40,000 courses on data and AI qualifications on various external learning platforms. We also started the Best Team initiative in IT because our greatest asset is our people. So it’s of great importance to us to attract and retain the best employees and enable them to fulfil their individual potential.

Does Mercedes only train the employees working in production or office workers as well?

We invest in the development of digital skills in all areas of the company. Regardless of whether they’re colleagues from production or administration, everyone needs the relevant knowledge and new skills to use AI applications effectively in their everyday work. In two pilot programs, we’re currently training more than 600 employees from all areas of the group to become data and AI specialists.

Do you expect generative AI will lead to job losses?

We can’t yet say with certainty what effects increased digitization and generative AI will have on future working life. What is clear, however, is working methods will change, as will job profiles themselves. That’s why qualification is the key to successful transformation.

How does the work of employees change in the course of the digital transformation process, and as part of the introduction of AI?

Some activities will certainly be able to transfer to AI applications in the future, like repetitive activities or those related to pattern recognition. But that’s something positive because it means more freedom for strategic or creative work is opened up, just as automation and production robots have changed how cars are made.

When we talk about training employees, how far has Mercedes-Benz progressed using generative AI?

We’re really productive with generative AI in some areas—not just talking about pilots. For example, we’ve been using GitHub Copilot in software development since May and we’re seeing significant gains in efficiency there. We also use generative AI in the customer environment. In Great Britain, for example, an intelligent virtual assistant can interact with customers on the website and give specific answers to questions about operating instructions and vehicle information. Also, in our data platform MO360, a digital ecosystem of production, a generative AI helps us analyze and process the data. And with the help of a large language model, the data, or data patterns, are available so they can be queried by production employees using natural language, not just by specialists using highly specialized database queries. We’re currently testing this using ChatGPT. Ultimately, AI accelerates a democratization of data use.

In which areas do you see the greatest potential for AI?

We’ve dealt with this question very intensively and analyzed both external studies and tried out AI internally. On one hand, there’s the software development. We see very significant increases in efficiency there, be it on the engineering and vehicle development side, or on the enterprise side.

Another area is the dialogue with the customer. For the foreseeable future, the direct interaction of the AI ​​with the customer, as is currently being tried out in the UK, will probably remain the exception. But I’m convinced AI applications will help further improve the customer experience and make processes more efficient.

One other area in which a lot of brain power must be invested is parametric design in engineering. There, AI will lead to major increases in productivity because it supports people at work.

And with the possibility of input by voice or keyboard—do employees have to be trained in the AI?

At the beginning, there’s initial training for the defined use cases in production. Plus, our employees have access to further training opportunities on the subject, including a learning path on prompt engineering. But they also learn how to use these tools in a creative way to try things out and see what works and what doesn’t.

In general, though, I believe prompting, or prompt engineering, is something you have to learn, so we’re considering whether we should offer training for it more widely throughout the company, and not just for selected IT and data professionals. It definitely helps to get more out of generative AI.

What teething troubles of AI or ChatGPT have you encountered so far? 

Hallucinations are certainly a challenge. That was also a very delicate balancing act in direct customer interaction in the UK. You can largely rule out hallucinations by plausibility checks and the associated restrictions, but if you set the criteria too narrowly, the machine will tell you, “I can’t comment on that,” more often than you’d like. You have to be very careful and find the right balance. How to get a grip on hallucinations is perhaps the most important question to be solved at the moment, which is also at the center of AI research.

Will Mercedes-Benz only train its AI tools on its own data?

Yes. For example, if we want to explain our vehicles to customers visually, then this can only be done with our own training data. Incidentally, the training takes place exclusively in secure areas of these AI environments, so the data can’t be made public. There’s also some public data we can use for AI, but especially in the production environment, we rely on our own data.

Apart from Azure OpenAI Services in the production environment, what roles do other AI solutions play for Mercedes-Benz?

OpenAI is currently being portrayed in the media as a bit of an AI spearhead. And there’s a very good technical solution too, but we won’t limit ourselves to that. Of course, other companies have interesting solutions. We’re starting to look closely at open source alternatives. In addition to the large proprietary providers such as OpenAI, Microsoft or Google, we need to understand the open source alternatives.

I also believe we shouldn’t think of AI as an engine that stands somewhere unto itself. It needs to be deeply woven into our systems and processes. That’s why we require all our system partners to use AI elements in their environments. It must find its way into the entire system landscape, and it will.