Menu
  • Home
  • Company

  • current page: Principles for green artificial intelligence

Company

Principles for green artificial intelligence

Artificial intelligence (AI) can contribute to climate protection. At the same time, the technology consumes a huge amount of energy and resources. It is therefore important to drive sustainable development and use of the technology. Our nine principles for green AI help us to do so.

The principles provide guidance on how AI solutions can be developed and used in a more environmentally sustainable way. And they show how risks - such as an increasing carbon footprint - can be countered at an early stage. We not only want to make the development and use of AI within Telekom more sustainable. We also want to set an impulse for others: for companies and institutions as well as for politics and science. The aim is to calibrate AI for sustainability right from the start.

Overview of Deutsche Telekom's 9 principles for green artificial intelligence.

Deutsche Telekom's nine principles for green artificial intelligence. © Deutsche Telekom

1. Green Electricity

Principle 1 for green artificial intelligence.

AI applications require more electricity than conventional IT applications. We focus on renewable energies throughout the entire value chain and encourage our partners to do the same. © Deutsche Telekom

2. Reusabiltiy in the value chain

Principle 2 for green artificial intelligence.

We promote multiple use of hardware, software, AI models, and data across our entire value chain. It allows us to operate more flexibily and efficiently and avoid unnecessary energy consumption. © Deutsche Telekom

3. Transparent CO2 Footprint

Principle 3 for green artificial intelligence.

Our AI development teams consider the CO2 emissions of hardware and software and analyze how changes to them affect the ecological footprint. © Deutsche Telekom

4. Dynamic Sizing

Principle 4 for green artificial intelligence.

Oversized hardware consumes unnecessary power. We adapt our software to our needs and intelligently scale down what we do not need for modelling, training or operation. © Deutsche Telekom

5. Optimized AI Models

Principle 5 for green artificial intelligence.

We use optimized AI architecture and AI models for specific use cases. We follow the modular principle and use software components multiple times where the task allows. This increases our efficiency and reduces energy consumption. © Deutsche Telekom

6. No Duplications

Principle 6 for green artificial intelligence.

We avoid duplications for similar use cases and find synergies. It prevents us from reinventing the wheel each time we develop software for new AI applications by creating transparency and enabling the reuse of code where possible. © Deutsche Telekom

7. Green Coding

Principle 7 for green artificial intelligence.

Codes differ in their power consuption. Because we want to develop in an energy-conscious way, we strive to code as efficient as possible. © Deutsche Telekom

8. Simplicity Matters

Principle 8 for green artificial intelligence.

We opt for AI models that are as simple as possible but still satisfy the use case. Once we have selected the right AI model we pay attention to the efficiency of the algorithms, to consume fewer resources. © Deutsche Telekom

9. End-to-End Responsibility

Principle 9 for green artificial intelligence.

We are responsible for the entire value chain and regularly check the carbon footprint of our AI applications' hardware and software. We stand by our end-to-end responsibility for Green AI with a special focus on Gen AI. © Deutsche Telekom


 

We foster the cooperative model. Get advantages out of a cooperative and complementary model of human-machine interactions.

Guidelines for Artificial Intelligence

These guidelines are self-binding. They define how we as DT want to deal with AI and how we will develop our AI-based products and services in the future.

Artificial Intelligence Guideline

FAQ