Stuttgart The world’s largest automotive supplier is progressing two years faster than planned in the application of artificial intelligence. “This year we will achieve our goal ahead of time that all products and solutions from Bosch contain AI or have been developed or produced with AI,” says Bosch CEO Stefan Hartung of the Handelsblatt.
Up to now, Bosch did not want to achieve this goal until 2025. The Swabians count themselves among the drivers of AI in German industry.
In detail, the achievement of goals is difficult to check from the outside. But it shows the handwriting of the new Bosch boss, who has been in office since the beginning of 2022. While predecessor Volkmar Denner, a physicist with a doctorate, opened up the foundation group to future topics such as AI and quantum technology and invested a lot of money in researching them, former McKinsey consultant Hartung relies more on the rapid implementation of the accumulated knowledge.
ChatGPT also plays a role: Of particular interest is a proprietary GPT system that has been trained on the basis of Bosch’s expertise. Such systems use artificial intelligence to understand human speech and thus generate a response similar to human speech.
In Bosch research, AI experts are currently exploring the possibilities offered by so-called foundation models. Up to now, Bosch knowledge has been stored in the company’s own database according to keywords. Only those who hit the keyword exactly have the chance of an optimal search result. “A solution with ChatGPT would search and find much more extensively and, above all, faster,” says Michael Fausten, who is responsible for AI research at Bosch.
The former McKinsey consultant relies more on the rapid implementation of the knowledge accumulated in the group.
(Photo: dpa)
The potential is great: With sales of more than 88 billion euros and more than 420,000 employees, Bosch is one of the German companies with the most patent applications. In 2022, the Swabians were in first place with 3946 patent applications at the German Patent and Trademark Office.
Patents, methods, development results, projects, training courses – for a group like Bosch, these are gigantic amounts of data that the company should be able to use more quickly and widely.
300 AI experts are swarming into the business areas
In total, more than 300 researchers at Bosch are working on the development and application of new AI methods. In addition, Bosch has now trained 26,500 associates in AI. The technology is now diffusing directly into the business areas. For example, the AI analysis platform developed by Bosch is already being used in more than 1,500 production lines.
Hartung is making a lot of steam to spread the knowledge more quickly across the group. He doesn’t stop at the favorite projects of his predecessor, who has placed AI development in a separate area.
>> Read also: Unusually clear criticism at Bosch: employees are demanding a future concept for all German locations
The specially established central AI division BCAI, which currently has 270 AI specialists, has not existed in this form as a separate organizational unit since last year. In fact, the experts are now also working directly on many of the Group’s projects.
“We want to implement even faster and bring AI into our applications. AI experts are active in research and also in our business areas. There is a close exchange,” emphasizes Fausten. However, the BCAI will be continued as a research community and will include all AI experts in Bosch research.
More software developers, but no dedicated AI campus
This also made another Denner project obsolete: In February, Bosch surprised when the much-announced AI campus in Tübingen, in which Bosch wanted to invest 100 million euros, was cancelled. One reason: With the pandemic, the need for office space in the group has fallen dramatically. According to Bosch, this is not a sign of a retreat from technology. On the contrary, the original goals are now being achieved more quickly, they say.
Even a previously solvent company like Bosch has to keep its money together when interest rates are rising sharply and banks and rating agencies are increasingly critical of the automotive supplier industry. As with all large suppliers, the costs for the transformation to electromobility eat up billions and hardly bring any money.
On the other hand, hiring software engineers on a massive scale is costing the company a lot of money. Their number increased last year by 6,000 to 44,000 in the group. Another 10,000 are to be added by the middle of the decade. This transformation to a software-defined company is also a tour de force. Bosch has to reinvent itself as an industrial group. Software and AI should be the key to this.
More on artificial intelligence:
According to Bosch, it has long been taking concerns about AI seriously. The company has set its own guidelines based on the EU’s AI Code. “We’re not interested in examining people in order to be able to predict their behavior,” says Fausten.
Machines that think for themselves should not replace human thinking, but complement it. Bosch uses AI for simulation in materials research. “We want to shorten development times in order to reduce the work on complex test benches. We bring AI into our products to make their behavior and usability better and smarter,” says Fausten. Video data cannot be analyzed without AI either.
Bosch is relatively unimpressed by Elon Musk’s recent criticism of unregulated AI. “We see artificial intelligence combined with the networking of products as a key technology and driver for the progress of our society. For us, it is central that the AI benefits people,” says Bosch boss Stefan Hartung, adding: “For years we have not only been dealing with the development of AI, but also with how to deal with it, with ethical and security issues .”
Bosch is developing AI that does justice to the company’s self-image of creating technology for life.
Smart ovens
Meanwhile, research on the Bosch campus continues unabated. The teams are working on anomaly detection with AI. It’s about traceability of chips in production, AI-based image processing that allows automated optical inspection of wafers and defect classification, and optimization of machine utilization in wafer production through material flow control.
The AI technologies should make the new chip factory in Dresden the most efficient semiconductor factory in the world. Intelligent cameras and sensors, for example, allow fire alarms to detect smoke development earlier. And in ovens, cameras use AI software to detect the desired degree of browning of roasts, cakes or pizza.
More: Data comparison through eye scan: ChatGPT creator Sam Altman wants to map humanity