SAP has created an AI ethics panel to guide its use of machine-learning technology. If only it had a similar committee for fraud allegations: it might have avoided the corruption scandal engulfing it in South Africa.
The German ERP giant – which is accused of kicking back $2m to secure state contracts – claimed it is the first European biz to create a external artificial intelligence ethics board: a five-person committee that includes technical experts and specialists in public policy, ethics, and bioethics. However, while several of them possess solid IT credentials, there's no one with a background in AI.
Rather, expertise in the evolving field will come from inside SAP. Specifically, the panel will work alongside the software house's internal technology ethics steering group – which includes people working in the field of machine learning as well as folks in corporate strategy, design, digital government, and data protection.
The ethics panel will dish out advice on how its specific use cases of AI technology may be affected by internationally accepted human rights principles and its own guiding principles for artificial intelligence.
SAP's seven AI principles
- We are driven by our values
- We design for people
- We enable businesses beyond bias
- We strive for transparency and integrity in all that we do
- We uphold quality and safety standards
- We place data protection and privacy at our core
- We engage with the wider societal challenges of AI
That seven-point plan aims to demonstrate the firm's commitment to do more than just what is legally required, and follows similar plans drawn up by Microsoft and Google.
"Bias can negatively impact AI software and, in turn, individuals and our customers," SAP's Corinna Machmeier explained on Tuesday. "This is particularly the case when there is a risk of causing discrimination or of unjustly impacting underrepresented groups."
To address this, SAP said technical teams will be required to "gain a deep understanding of the business problems they are trying to solve," and think hard about the quality of data required. The biz is also investigating technical ways to mitigate biases, as well as increase the diversity and interdisciplinarity of its techies. The wider the backgrounds of its developers, the better it will be at catching social and cultural biases in training data and algorithms, we assume.
The software house is also committed to ensuring its AI code undergoes rigorous testing in the real world, and making the input, capabilities, intended purpose, and limitations of its systems clear to customers so no one is duped by hype – or tries to use the software for naughty purposes.
"We recognise that, like with any technology, there is scope for AI to be used in ways that are not aligned with these guiding principles and the operational guidelines we are developing," Machmeier added. "Where there is a conflict with our principles, we will endeavour to prevent the inappropriate use of our technology."
The firm's final commitment is to look beyond its own technology, and tag along with existing efforts to identify and mitigate increased use of AI in other industries and disciplines.
For instance, the European Commission established a high-level expert group on AI earlier this year, of which Markus Noga, SAP's senior veep for machine learning, is a member. This group is working to draw up a European AI strategy by early 2019, which SAP will no doubt take an interest in.
The seven principles will, essentially, be used to assess SAP's products – which currently have to pass 150 quality checks – and will be developed and added to by the two steering groups.
Meanwhile, SAP is taking its ethical approach to AI in a mildly novel direction with the launch of what it is branding a "Plastics Cloud". The biz, along with the Design Thinkers Academy in London, England, and various NGOs, is collecting existing and live data from the plastics supply chain to plug into its SAP Leonardo toolkit in a bid to identify new ways to reduce waste. ®