Responsible and Ethical Use of Artificial Intelligence at Porsche
The adoption of Artificial Intelligence (AI) technologies at Porsche is progressing at an ever greater pace, with more and more use-cases being developed or already put to production. This trend affects both our customers and employees, with promises towards more personal and engaging user experiences, as well as more efficient company processes with a high degree of automation.
For instance, a report from the World Economic Forum (WEF) estimates that by 2025, more than half of all current workplace tasks will be performed by machines as opposed to 29% today. With that said, the speed at which artificial intelligence advances creates a pressing need for debate and reflection. When my colleague Petra and I first discussed the idea of open debate within Porsche about this topic, we took inspiration from the German TV show “Hart aber Fair”.
After a few weeks of planning and collecting topics within our AI community, we invited six AI and IT experts with diverse backgrounds. We also prepared little TV clips and so-called “fact checks” around bias, data privacy and more to give structure to the discussion:
Meet the experts
Akan Isik is a member of the Workers Council Zuffenhausen. He argued that a loss of jobs caused by the technological change must be avoided at all costs. Jörg Sädtler has founded multiple startups, including predict.io, which first targeted a crowd-based parking app. With more and more data being acquired from their customers and increasing ethical concerns, Jörg and his team decided to shut down their startup with a heavy heart.
While Peter Schaar, former German Federal Commissioner for Data Protection and Freedom of Information, said that “artificial intelligence is both an opportunity and a challenge” and that “the design and use of AI systems must comply with our ethical values and legal requirements”. Director Data Protection & Privacy at Porsche, Christian Völkel, stated that “digital self-determination does not have to run counter to business interests. A high degree of transparency is necessary to convince the customer of our business models.”
Anja Hendel, Director of the Porsche Digital Lab in Berlin, explained that we require operational methods for the development of fair and responsible AI. Singular measures are simply not sufficient. Mattias Ulbrich, CIO at Porsche, said that he expects artificial intelligence to benefit business sectors and the overall economy, highlighting that it can replace human labour in routine tasks and provide benefit and greater freedom for everyone.
AI and the future of work
There has been widespread speculation about the impact of artificial intelligence on work and employment. The advent of AI has fueled fears that the majority of jobs are at risk of being automated. However, if done right, AI can help people excel at their jobs, freeing up time for creativity and innovative thinking. We have to make sure that humans will remain in the lead and that artificial intelligence is used to augment, not replace human talent. Akan Isik made clear that it is necessary to continuously develop our workforce by providing training opportunities for our employees.
Building fair algorithms
Much discussed in the press, the bias in AI recruiting tools has led to the systematic exclusion of women or minorities. For example, a tech company’s machine learning model taught itself that female candidates are not suitable for the job because the training data mostly contained men. This lead to penalizing résumés that included the word “women” and preferring male candidates instead. Although the tool was quickly abandoned, Peter Schaar lined out that it taught us an important lesson regarding the limitations of artificial intelligence and machine learning. Without representative training data, there will be no representative output.
Therefore, the participants expressed a widespread consent that explainability and transparency is crucial for the use of AI algorithms at Porsche. Also, assessing the socioeconomic impact of AI will be necessary for developing sound policies and good AI solutions.
Technologies for Responsible AI: Synthetic Data and Trusted AI
We also discussed technologies that can help to excel in the development of ethical and responsible AI. Synthetic data, for example, is computer-generated data mimicking real data. It is derived from real data and exhibits the same statistical properties and relations with provable guarantees towards privacy and data security. Thus, it cannot be traced back to any individual. Similar methods like differential privacy can further be used to provide guarantees for user privacy.
In addition, trusted computing can help to securely execute AI algorithms on very sensitive data. Therefore, a processor creates a so-called enclave, in which all data is fully encrypted and the algorithm running can be fully verified in advance to sharing the data with it. Technologies, however, can only be one part of the equation, processes and people play a similarly important role.
Conclusion
AI is and will be a fundamental part of Porsche’s strategy for the future. The fruitful debate motivated us to further drive open debates and transparent discussions in the future. Finally, there is a broad consent within Porsche and our community to excel ethical and responsible AI for the benefit of our customers and employees.
Dr. Petra Beenken is IT Architect at Porsche, Tobias Grosse-Puppendahl ist reponsible for Platform Incubation Emerging Technologies at Porsche. Follow us on Twitter (Porsche Digital, Next Visions), Instagram (Porsche Digital Lab Berlin, Porsche Digital, Next Visions) and LinkedIn (Porsche Digital Lab Berlin, Porsche Digital) for more.
