Artificial intelligence is more than a technology
AI’s impact - both positive and negative - should not be underestimated. It may still seem far removed from many people’s daily lives, but you use it almost every day: in navigation systems, Google Search, product recommendations, or the app that tells you which bus or train to take. Deutsche Telekom also uses AI - for example, in customer service as a chatbot, at T-Systems to support maintenance, and in products such as the Connect App and the upcoming Smart Speaker.
Why does Telekom have ethical principles for AI?
Like all new technologies, AI brings both opportunities and challenges. Consider how algorithms can adopt human biases, or whether a chatbot should be recognizable as such. There is also the black box problem (you enter data and receive a result, but do not know how the AI system arrived at it).
Because it is important to use AI thoughtfully, Telekom developed guidelines for the ethical use of AI as early as 2018.
What are the guidelines about?
These guidelines are self-binding policies tailored to our business models. They define how we at Deutsche Telekom want to approach AI and how we will develop our AI-based products and services in the future. The basic idea is that intelligent algorithms, robots, and computers are initially just tools and are neutral in themselves. It is therefore up to us to use them in positive ways without overlooking the risks and to handle those risks responsibly.
We explored how to ensure this positive use not only internally, but also through numerous discussions with other companies, experts, and institutions in the field of AI in Germany, the United States, and Israel. Examples include Facebook, Microsoft, Amazon, Google, Stanford, Open AI Initiative, Allen Institute, and Partnership on AI. We are proud to be one of the first companies worldwide to establish such self-binding guardrails/principles for action.
Our approach: Nine principles for AI ethics
These guidelines are guided by the principle that artificial intelligence should support our actions. Technology should help people and expand their capabilities. Selected examples include:
- We take responsibility.
A clear definition of who is responsible for each AI system. - We are committed to transparency.
Transparency when a customer communicates with an AI system, and transparency about how customer data is used. - We provide security.
Our customer data is protected from unauthorized external access (data security and data protection). - We stay in control.
We remain ready at all times to intervene in our AI systems to prevent or reduce harm.
These are just a few examples. You can read all nine guidelines here.
We do not claim that our principles are some kind of silver bullet. Instead, we want to continue developing the Guidelines through dialogue and build them on an even broader consensus.
