Emerging themes

In our 2018 October update we explored the developing regulatory approaches to “big data” and artificial intelligence. Since then there has been an intensification of activity in this space around the world and it is helpful to look at the themes emerging. (We have also reported in detail separately certain of these developments and links are given to those reports).

A number of themes weave consistently through recent developments:

  • at a practical level, regulators and policy makers are working hard to skill up. Links to experts and academia are being forged. There is a focus in particular on tools and methods to unpick processes and understand outcomes so as to enable adverse and unintended consequences to be identified;
  • both regulators and industry see a need for widely accepted, ethical principles to encompass the design, operation and outcomes of automated decision-making and other data driven processes. Decisions around the type and source of data that may appropriately be used, and acceptable purposes for which data may be used, cannot be ducked and are challenging. There will be clear business and reputational risk for those found to be on the wrong side of any emerging consensus;
  • the need for global cooperation at a regulatory level to avoid a “race to the bottom” is front of mind with many regulators;
  • regulators are placing a lot of stress on human oversight, accountability and governance as a linchpin of envisaged systems of regulation and ethics codes; and
  • impact on the vulnerable and marginalised is a key focus. Market participants should expect to be required to take due care to anticipate and check the impact of their systems on the vulnerable and minorities and if they fail to do so, to be held to account for unintended consequences they should have foreseen.

UK developments

In July the Financial Conduct Authority (FCA) announced that it would be collaborating with the Alan Turing Institute in a year-long joint project designed to investigate the use of AI in the financial services sector. The FCA wants to deepen its understanding in this area, with a focus on specific harms and safeguards.

The Centre for Data Ethics and Innovation (CDEI), an independent advisory body, led by a board of experts, tasked by the UK Government to advise on data driven technologies, launched reviews into on-line targeting (the use of data driven technologies to target content, products and services to different individuals) and algorithmic bias in the spring and has recently published interim reports of its findings. Click here and here for our review of the CDEI interim reports.

EU developments

In our January Update we commented on the draft guidelines for Trustworthy Artificial Intelligence developed by the EU’s High Level Expert Group on Artificial Intelligence (the HLEG and the Guidelines, respectively). Following feedback a revised version of the Guidelines was published in late June andin July the European Commission launched a pilot phase for testing their revised ethics checklist. Click here for our review of the revised Guidelines and the pilot process.

HLEG has also set out broader policy and investment recommendations in the context of artificial intelligence. Click here for our report on the main take-aways.

Other global developments

In May the members of the OECD and a number of other states adopted the OECD AI principles.

In June most of the G20 (excluding Russia and China) adopted the G20 AI Principles, building further on the OECD AI principles.

In May the Beijing AI Principles was jointly unveiled Saturday by the Beijing Academy of Artificial Intelligence (BAAI), Peking University, Tsinghua University, Institute of Automation and Institute of Computing Technology in Chinese Academy of Sciences, and an AI industrial league involving firms like Baidu, Alibaba and Tencent. This was followed in June with the publication of Governance Principles for the New Generation Artificial Intelligence by China’s National Governance Committee for the New Generation Artificial Intelligence.

Although each of these is much less granular than the EU Guidelines, and there are differences of emphasis, there is significant overlay in the broad themes covered.