Bosch draws up AI ethical code
February 19, 2020
Bosch has established ethical red lines for the use of artificial intelligence (AI). The German technology company has issued guidelines governing the use of AI in its intelligent products.
The AI code of ethics is based on the maxim that humans should be the ultimate arbiter of any AI-based decisions.
“Artificial intelligence should serve people,” said Bosch CEO Volkmar Denner at this week’s Bosch Connected World (BCW), the company’s annual IoT conference in Berlin. “Our AI code of ethics provides our associates with clear guidance regarding the development of intelligent products. Our goal is that people should trust our AI-based products.”
AI is a technology of vital importance for Bosch. By 2025, the aim is for all its products either to contain AI or have been developed or manufactured with its help. The company wants its AI-based products to be safe, robust and explainable.
“If AI is a black box, then people won’t trust it,” said Michael Bolle, Bosch CDO and CTO. “In a connected world, however, trust will be essential. Bosch is aiming to produce AI-based products that are trustworthy. The code of ethics is based on Bosch’s invented-for-life ethos, which combines a quest for innovation with a sense of social responsibility. Over the next two years, Bosch plans to train 20,000 of its associates in the use of AI. Bosch’s AI code of ethics governing the responsible use of this technology will be part of this training programme.”
AI technology can help overcome problems such as the need for climate action and optimise outcomes in a host of areas such as transportation, medicine and agriculture. By analysing huge volumes of data, algorithms are able to reason and make decisions. Well in advance of the introduction of binding EU standards, Bosch has therefore taken the decision to engage with the ethical questions that the use of this technology raises. The moral foundation for this process is provided by the values enshrined in the Universal Declaration of Human Rights.
The code of ethics stipulates that AI should not make any decisions about humans without this process being subject to some form of human oversight. Instead, AI should serve people as a tool.
Three possible approaches are described. All have the following in common: in AI-based products developed by Bosch, humans should retain control over any decisions the technology makes.
In the first approach (human-in-command), AI is purely an aid, for example in decision-supporting applications, where AI can help people classify items such as objects or organisms.
In the second approach (human-in-the-loop), an intelligent system autonomously makes decisions that humans can, however, override at any time. Examples of this include partially automated driving, where the human driver can directly intervene in the decisions of, say, a parking assistance system.
The third approach (human-on-the-loop) concerns intelligent technology such as emergency braking systems. Here, engineers define certain parameters during the development process. There is no scope for human intervention in the decision-making process itself. The parameters provide the basis on which AI decides whether to activate the system or not. Engineers retrospectively test whether the system has remained within the defined parameters. If necessary, these parameters can be adjusted.
Bosch also hopes its code of ethics will contribute to the public debate on AI.
“AI will change every aspect of our lives,” Denner said. “For this reason, such a debate is vital.”
It will take more than just technical know-how to establish trust in intelligent systems; there is also a need for close dialogue among policy makers, the scientific community and the general public. This is why Bosch has signed up to the High-Level Expert Group on Artificial Intelligence, a body appointed by the European Commission to examine issues such as the ethical dimension of AI.
In a global network currently comprising seven locations, and in collaboration with the University of Amsterdam and Carnegie Mellon University, Bosch is working to develop AI applications that are safer and more trustworthy. Similarly, as a founding member of the Cyber Valley research alliance in Baden-Württemberg, Bosch is investing €100m in the construction of an AI campus, where 700 of its own experts will soon be working side by side with external researchers and start-up associates.
Finally, the Digital Trust Forum, a committee established by Bosch, aims to foster close dialogue among experts from international associations and organisations.
“Our shared objective is to make the internet of things safe and trustworthy,” Bolle said.