Swiss Government publishes AI Guidelines

Bundeshaus - Swiss Government publishes AI Guidelines

Written by:
Kevin Schawinski (CEO at Modulos)

Federal Palace of Switzerland (Bundeshaus).

The Swiss government has published its guidelines for the use of AI within the federal government. With these guidelines, the Swiss government puts trust, openness and security at the heart of all AI systems which government agencies plan to use.

At the center of efforts around AI, the Swiss government rightfully puts humans. But, to demystify AI and to make it more accessible to non-experts as well as for experts it needs to be abstracted and simplified. Further, it should be transparent, explainable, and free of any bias. 

A consequence of this is that all required user choices are made explicit and the impact is understandable to the user. This is one of the driving philosophies, when building and expanding Modulos AutoML. AutoML will enable the user to build AI systems compliant with the guidelines by the Swiss government for any industry, but, especially for those so far not considered AI typical such as farming.

One of the characteristics of AutoML tools is furthermore that selecting and training the best AI models is done by the core engines instead of leaving it to the user. This allows for a systematic search of the space of AI models and their parameters, which is transparent and can be audited, and removes the potential for unwanted biases. Such biases could lead to unintended consequences for individuals and may result in the AI application not meeting the guidelines from the Swiss government.

In addition, all AI Solutions (aka trained Machine Learning models) generated by Modulos AutoML are completely accessible to the user. All model code, model parameters and neural network weights can be analyzed and tuned without restrictions. This allows expert machine learning users to generate a stream of prototype models efficiently followed by using their expert knowledge to further optimize if needed. 

All solutions can be deployed without artificial limitations and restrictions. And, the AutoML platform itself is secure and safe, protecting personal data through its many deployment options including air-gapped systems or on premise deployment. This allows our customers to get access to cutting edge machine learning while ensuring to stay compliant with GDPR/DSGVO and other data security mandates.

Get in touch with us and one of our experts will be reaching out to you for a conversation on AutoML and related topics.

And if you want to read the entire guidelines, here they are: