Florian Michahelles

Head of Research Group, Artificial & Human Intelligence

"Don’t be afraid of AI – how digital companions can helps us shape a better future of work"

Interview with Florian Michahelles
January 14, 2020

About Florian

We focus on the creation of Digital Companions for Industry. A Digital Companion is an entity that enhances human capabilities. It digests, integrates, and shares information with humans so they can focus on meaningful tasks. According to the user needs, a Digital Companion can act as a guardian, assistant, or partner. The Digital Companion embodiment adapts to what best suits the current user context and needs.
To make our Digital Companion vision a reality, we specialize in Semantic Technologies, AI, and innovative user interfaces, such as Augmented and Virtual Reality, and voice user interfaces.
Our quest for innovation is not limited to already commercialized technologies. We need to shape the future. Our efforts also include research in Quantum Computing in conjunction with distinguished partners such as University of California, Berkeley and NASA.
Before, as associate director and research coordinator of the Auto‐ID Labs (autoidlabs.org) and lecturer at ETH Zurich, Florian concentrated on developing new service concepts in close collaboration with partners from industry. His central research interest was the conception, development and evaluation of the electronic integration of daily‐life objects and the connected, independent exchange of information between these objects (internet of things).

we.CONECT: The pop in your job – you work as Head of Research Group, Artificial & Human Intelligence at Siemens – which tasks would you highlight as particularly interesting? What is most exciting about your role, and why are you passionate about your job?

Florian Michahelles: As most exciting I perceive working on cutting edge technologies, engage with thought-leaders, and translate and transfer results to practice. I’m passionate about having one step further into the future with the power of influencing directions.

we.CONECT: Your research area is focusing on Hybrid Intelligence and how digital companions can help us shape a better future of work. You will present this topic at Industry of Things World USA. Can you already share some insights on how AI can support and augment human capabilities and how customers and companies can benefit from human centered systems?

Florian Michahelles: We focus on how AI und human intelligence can complement it each other. Despite the hype of AI, AI can yield amazing results where data of the past represents possible actions of the future. Then, AI can create the _illusion_ of intelligence where is in reality it is just about pattern recognition and machine learning. Our approach of digital companion recognizes this by providing visualizations and interaction paradigms allowing a more smooth handover between AI and human user, such that the human user can build upon the AI illusion and add the human power of problem solving, critical thinking, and social collaboration.

we.CONECT: You studied Computer Science at LMU, MIT. What were the most important drivers for you to focus on Human Intelligence and participative design?

Florian Michahelles: When I studied CS at LMU the study field of HCI did not exist there. However, I was in the lucky situation of being able to choose experimental psychology as a minor. This way I got to experience both the system world of information technology and theories and experimentation about human perception, memory, and information processing. Later, during my PhD studies at ETH Zurich I got finally exposed to the practices of user-centered and participatory design.

we.CONECT: What will be the biggest challenges for companies to reduce reservations against AI in daily operations?

Florian Michahelles: There is no magic. I see the challenge two-fold: 1) ignoring AI due to a lack of curiosity and dominance of classical mechanical engineering stereo-typically control-loop thinking. 2) overhyping AI as the holy grail of unveiling unexpected insights into data as never seen before. I strongly believe it is important to understand both opportunities but also limitations of AI as statistical and mathematical methods capable of analyzing highly multi-dimensional data. And yet, results will be only as good as the quality of the available data and the questions being investigated. It is only magic if you don’t understand it.

we.CONECT: Who should be involved in the decision process of integrating more AI in companies? Should there be a committee, a “Chief Ethical Officer” or do you think those decisions will be slowly taken with new generations of employees?

Florian Michahelles: First of all, you should integrate today’s domain experts: how do they make decisions today? Which parameters do they include into their rationale? Which are their decision guidelines, rules and thresholds? What are the sweetspots, i.e. tacit knowledge, often referred to as experience/intuition/gut feeling? How do they correct themselves? Has the rationale changed over-time? What’s the available historic data?
Secondly, data-scientists should join the conversation and experiment with modeling the rationale of the domain experts, compare results, iterate, distill opportunities and limitations. As long as there is transparency about which data AI is building upon, and which conclusions are derived, I don’t believe an ethical officer is needed other than complying with today’s policies of compliance and legal frameworks.

we.CONECT: Industry of Things World USA is a highly interactive networking event – which conversations are you particularly looking forward to?

Florian Michahelles: I’m particularly looking forward to conversations on dealing with limitations of AI by providing seamless hand-overs to human experts.

View Session