Ericsson on AI ethics

AI – ethics inside? Challenges and opportunities for the future
Ericsson IndustryLab, 2021
Download the report

This report by the IndustryLab of Ericsson, the Swedish multinational, aims to introduce the ethics of AI and explore how this fast-growing technology needs to align with humans’ moral and ethical principles if it is to be embraced by society at large.

Using qualitative data from a range of experts (all interviews conducted between January and March 2021) and considering well-established philosophical examples, the paper focuses on the unintended consequences of AI usage and how human error or oversight could cause severe and lasting damage, despite the many opportunities it creates.

It also explores other quandaries such as the inherent biases within AI systems, how the successful implementation of AI relies on building consumer trust, and how legislation used to regulate AI needs to be tailored to each individual industry. Upcoming editions will then dig deeper into the various opportunities and challenges that AI brings to different sectors.

Key findings

  1. Lack of focus on unintended consequences of AI usage
    Much of the AI research being conducted today focuses on ways to mitigate intended misuse of the technology. However, the experts interviewed generally agreed that the unintended consequences of normal deployment could be equally damaging and therefore need more attention.
  2. The impossible challenge with bias in AI
    Left unchecked, biased data could risk furthering gender inequalities or fueling continued racial injustice. Indeed, several of the experts agree that striving for unbiased AI is virtually impossible, though they added that these effects could be minimized.
  3. The paradox of trust
    AI is often used to support human decision making so it is important to build up trust in these systems, but without fostering an overreliance on the technology.
  4. AI ethics is important for everyone
    AI ethics is not just important for data scientists – it’s crucial for everyone. AI will soon permeate many facets of people’s work and personal lives.
  5. Today’s AI guidelines are not enough
    To better support AI development for business, industry and society, today’s guidelines, rules and regulations need to be improved to cover more than just consumer-focused variables.
  6. Ethics by design is key for acceptance
    While technology itself can never be ethical, it can be ethically aligned. An ethical dimension should run parallel to areas like security and privacy within the design process – addressing these problems early will save time and money in the long run.

Experts

  • Victor Bernhardtz, Ombudsman for Digital Labour Markets, Unionen
  • Jacob Dexe, Research Institutes of Sweden
  • Virginia Dignum, Professor of Ethical and Social Artificial Intelligence at Umeå University, member of EC HLEG on AI
  • Andreas Henningsson, Domain Architect – Cognitive Data Analysis & Innovation, Swedish Social Insurance Agency
  • Isaiah Hull, Senior Economist, Central Bank of Sweden
  • Magnus Kjellberg, Section leader Digital R&DI, Sahlgrenska University Hospital
  • Sofia Löfstrand, Senior Project Manager, Drive Sweden
  • Inese Podgaiska, Secretary General, Association of Nordic Engineers, ANE
  • Amritpal Singh, Founder and CEO, Viscando
  • Mikko Viitaila, Regional Technology Officer for Microsoft Western Europe