Intro

Creating product experiences that all people love has always been our compass. As we integrate generative AI into the experiences we craft, we must stay focused on harnessing its potential for good. As these technologies grow ever more powerful, those of us who build with it have an important duty – to put ethics before profits or any other priorities. We all play a crucial role in advocating for responsible innovation. You should always refer to our SAP’s AI Ethics Policy Handbook to learn more about AI ethics and our risk framework.

When it comes to sustainability, we must take a holistic approach that considers the environmental, social, and economic impacts of what we design.

Why It Matters

Generative AI has the potential to empower people who use our products in a way that creates real customer value – but we need to mitigate any potential harms to society, the environment, and humanity.

Here are four aspects:

• Energy consumption:
  Generative AI models use a lot of energy. This can have a significant environmental impact, since the energy used to power these models comes from fossil fuels.

Estimated annual emissions from large-scale adoption of generative AI, by lifecycle stage, in annual CO2 emissions

• Data usage:
  A ton of datasets are used to train large language models (LLMs) like those being used for generative AI. This data can come from a variety of sources, and it’s important to consider the environmental impact and potential harm of collecting and storing this data.
• Bias:
  AI models often reflect and even amplify the biases in the data they are trained on. This has harmful and often severe consequences, so it’s critical that we take steps to mitigate bias in generative AI models.
• Responsible use:
  It’s important to use generative AI in a responsible way, and to be aware of the potential risks and unintended outcomes. We need to question whether or not generative AI is the best way to solve the problem that we are trying to solve.

What We Can Do

• Take our responsibility for data seriously
  Generative AI experiences should be designed in a way that’s inclusive and empowering to a diverse range of people across the globe. We must mitigate the risk of harmful bias in the data that is used to train the generative AI models by incorporating activities like those often found in “red teaming” into our product development process. It’s on all of us to make sure that the data we want to use in these generative AI models is anonymized and cleaned of any personal information to the best of our abilities to curb bias and discrimination. This will help us reduce potential harm and make sure that our product experiences do not discriminate against or exclude any group of people.
• Choose the right model
  Not all generative AI models are created equal, so it’s important to choose one that’s as efficient as possible for your needs. Regardless of your role, it’s critical that we only use generative AI to solve problems that it is uniquely suited to solve.
• Optimize the model
  Once you’ve chosen a model, you can optimize it to reduce its environmental impact. This can be done by adjusting the parameters of the model or by leveraging prompt training and fine-tuning existing generative AI models on our specific domains.
• Fine-tune the model
  In some cases, it may be possible to fine-tune a generative AI model to make it more efficient and lessen its environmental impact. We can actively partner with data scientists and AI engineers to help identify the areas where a model can be fine-tuned to improve its sustainability.
• Measure, monitor, and improve
  It’s also vital that we continuously monitor generative AI-powered experiences after launch – to protect our users and customers from the unintended consequences of hallucination and harmful bias, and to ensure data privacy and protection.

We all have the power to create generative AI-powered experiences that are good for the planet and good for people. Use your skills to design experiences that are efficient, use less data, and are accessible to everyone.

You can make a real difference in the world.

References

Image: Estimated annual emissions from large-scale adoption of generative AI. Published by Kasper Groes Albin Ludvigsen, in ‘Towards Data Science’ on Medium.