Share this post:
Today the independent EU High Level Expert Group on AI (HLEG) issued policy and investment recommendations for trustworthy AI. Francesca Rossi, Global AI Ethics Leader at IBM and a member of the EU HLEG, provides highlights below.
Set up by the European Commission in 2018, the HLEG is composed of a broad spectrum of AI stakeholders, and was mandated to develop guidelines and policies for a European AI strategy. Today’s recommendations on AI policy and investment are the second major step in that effort. They follow the group’s AI Ethics guidelines issued in April that set a new global standard for efforts to advance AI that is trustworthy and responsible.
As with the AI Ethics Guidelines, today’s recommendations are the result of thorough discussions within the HLEG and provide a comprehensive blueprint for developing a thriving AI ecosystem in Europe that can have a positive impact across the world.
The 33 recommendations are timely as governments around the world seek input and guidance to define their own AI strategies. Highlights from these recommendations include:
- While concerns about AI must be addressed by governments and companies, it is clear from the group’s recommendations that an immediate rush towards broad-ranging regulation is not the best way to achieve trustworthy AI while supporting innovation and competitiveness. The recommendations state that unnecessarily prescriptive regulation should be avoided in contexts characterized by rapid technological change, where it is often preferable to adopt a principled-based approach, as well as outcome-based policies, subject to appropriate monitoring and enforcement.
- Policy makers should consider the different contexts of AI – whether B2C, B2B or P2C. This is an important differentiation because – as stated in the HLEG recommendations – not all risks are equal. For example, the risk of a power imbalance between consumers and dominant companies in the B2C space could be greater than from negotiated agreements between parties in the B2B space. Therefore, government policies and regulation should indeed take a risk-based precision-driven approach – adapting to the context and targeting only where the problem is.
- A thorough mapping and evaluation of all existing AI-relevant EU laws is recommended. The goals of this exercise are, amongst others, to identity existing legislation that already fulfils the AI policy and investment goals and drives AI ethics. This initiative will be very useful in shaping clear AI policies and avoiding multiple and/or contradictory pieces of legislation. Also, in this spirit is the recommendation to avoid cumulative regulatory interventions at the national level so we create a truly Single Market for Trustworthy AI.
- Placing the human at the centre of AI was at the core of the AI Ethics guidelines and it rightly continues through the policy and investment recommendations. For IBM, a key platform of human-centric AI is ensuring that all sectors of the population have the skills to benefit from AI. To achieve this, the HLEG rightfully recommends a redesign of our education system from pre-school to higher education. One of the solutions IBM proposes is P-TECH, an education programme we developed with educators that provides young people – the majority from disadvantaged communities – with the qualifications and professional skills they need to be ready for the modern workforce. Well established in the US, Australia, Morocco and Taiwan, P-TECH is now underway in Ireland, the UK, France, the Czech Republic and Italy, involving innovative partnerships between local schools and local and international companies. Also, SkillsBuild is a new IBM digital platform in partnership with NGOs that provides job seekers with the coaching and learning they needs to gain the skills to re-enter the workforce. Following its launch in France, the program will expand to Germany and the UK in 2019.
- It is significant that the HLEG recommends that the public sector, including governments, serves as a catalyst for the update and scaling of Trustworthy AI. This is an important route to expand access to and familiarity with the technology among the individuals that governments serve.
- With significant AI research capabilities in Europe and with active participation in Horizon 2020 projects, IBM strongly endorses the recommendations to strengthen and unite Europe’s research capabilities and to harness an open and innovative investment environment. Openness in all its guises will drive AI, including in the future Horizon Europe and Digital Europe programmes. This is why international collaboration is essential for research: societal and scientific challenges can rarely be addressed by any one organization or region.
A final word on the approach to developing the recommendations. In my 30-year career as an AI researcher and in my work in different AI organizations across the world, I have seen the enormous benefits that collaboration brings. This now also applies to efforts underway in capitals worldwide to develop AI policy. IBM has just responded to the United States Department of Commerce’s National Institute of Standards and Technology (NIST) consultation regarding the development of standards and tools to build and advance trusted AI. We were proud to have contributed to the recent OECD Principles on AI. And, using our experience as a global leader in the advancement of trusted AI, we recently presented 10 AI policy priorities for governments and public policy makers.
The European Commission approach to developing a Trustworthy AI framework is a live example of how a multilateral approach can bring successful results. The HLEG brings together not just technology experts but representatives of many different sectors including multiple academic fields, industries, human and consumer rights associations. I applaud the European Commission for instigating this process which has resulted in recommendations that are both ambitious and feasible. We now look forward to the new European Commission using the recommendations as the basis for its AI and Digital Strategy 2020-2024.
-Francesca Rossi, AI Global Ethics Leader, IBM