e-Legal Working Group AI and Machine Learning

23 May 2019

We want to take this opportunity to extend a special thanks to Caitlin St. Vincent Welch, from Swiss Re for helping to organise the inaugural e-Legal working group and Rackspace International for hosting.

It was great to see so many experienced legal professionals from both sides, in-house and private practice lawyers all gathered to share knowledge, experience and discuss hot topics impacting their roles every day.

The first presentation was dedicated to AI and Machine learning and its legal considerations, presented by Dr Daniel Ronziani and Caitlin St Vincent Welch. The introduction covered a brief definition of AI and some uses but the focus turned towards some legal considerations, such as intellectual property, criminal liability and contractual liability. 

So, what is AI and Machine learning, Daniel explained:

"It can be looked at through a human lens as AI systems are intelligent and independent, which are also human attributes. AI combines the two factors symbolic learning and machine learning. Symbolic learning is the ability to perceive the environment and to judge (computer vision) and the ability to act (robotics). From a human perspective, it should be compared to the fact that our brain contains around 80 to 100 billion cells and many links connect all of these cells. Deep learning summarises the quality of the brain to construe new connections. To improve machine learning, the availability of large amounts of data is crucial — neural networks bringing an essential shift from image recognition to pattern recognition."

AI is already an active part of our everyday lives, through chatbots, virtual assistants and processing for insurance claims. Using the example of chatbots, and the frequency in which they are being used the quality of service is being greatly improved. This gives customers 24/7 support and for the businesses at a lower cost. Virtual assistants can help to increase productivity by supporting users in scheduling their daily agenda. They learn from the user's past interactions and can come up with accurate results that match user preferences.

Several challenges that we briefly discussed and require a more dedicated session included: Black box AI, the biggest challenge faced would be the regulators demand that there is an ability to explain the decision process. What will be the trade-off for developers, and will it slow down the innovation process?

Whilst on the topic another challenge, we are facing is bias and discrimination. As the quality of an AI system is relying on the input data that developers are feeding it. Moreover, the data they input feeds the system can never be entirely neutral but will always contain conscious and unconscious bias.

Another question is turning around Intellectual Property. As systems can produce an output, who will be the owner of it? Is it going back to copyright? Copyright laws are only protecting outputs that were created by humans, machines. Machines at the moment have no performance protection. For companies, this brings the risk that they invest money in R&D done by AI, but their IP is not protected. Does the law need to be changed?


Chair: Philippe Lucet, VP General Counsel R&D and IP at Nestle / Netcomm Suisse Board Member.

Keep in touch

The first and only
Swiss Association of



Centro Galleria 1 A
Via Cantonale
6928 Manno CH


T. +41 (0) 91 210 47 36

VAT no.: CHE-485.124.256 IVA


T. +41 (0) 44 500 21 73
@ digitalswitzerland


Blue Box, Chemin du Pré-Fleuri 3
1228 Plan-les-Ouates CH
T. +41 (0) 22 510 71 03
@ Fongit Foundation