Reports  |    |  February 4, 2020

CDEI Review of Online Targeting: Final Report and Recommendations

Report published by the UK Centre for Data Ethics and Innovation. 133 pages. Independent advisors on AI and data-driven technology publish first recommendations to government on social media targeting.

Executive Summary:

  1. Data-driven online targeting is a new and powerful application of technology. Using machine learning, online targeting systems predict what content is most likely to interest people, and influence people to behave in a particular way.
  2. Personalisation of users’ online experiences increases the useability of many aspects of the internet. It makes it easier for people to navigate an online world that otherwise contains an overwhelming volume of information. Without automated online targeting systems, many of the online services people have come to rely on would become harder to use.
  3. Online targeting systems are used to promote content in social media feeds, recommend videos, target adverts, and personalise search engine results. Online targeting is already an important driver of economic value and is a core element of the business models of some of the world’s biggest companies. It enables individuals and organisations to find a bigger audience for their stories or point-of-view, and businesses to find new customers. Automated systems now make decisions about a significant proportion of the information seen by people online.
  4. As the underlying technology continues to develop, online targeting will continue to grow in sophistication and it will be used in novel ways and for new purposes. There are already a number of services that help people make positive changes by tracking their health, diet or finances and use personalised information or nudges to influence their actions. There is significant potential for further innovation.
  5. However, online targeting systems too often operate without sufficient transparency and accountability. The use of online targeting systems falls short of the OECD human-centred principles on AI (to which the UK has subscribed), which set standards for the ethical use of technology. Online targeting has been blamed for a number of harms. These include the erosion of autonomy and the exploitation of people’s vulnerabilities; potentially undermining democracy and society; and increased discrimination. The evidence for these claims is contested, but they have become prominent in public debate about the role of the internet and social media in society.
  6. Online targeting has helped to put a handful of global online platform businesses in positions of enormous power to predict and influence behaviour. However, current mechanisms to hold them to account are inadequate. We have reviewed the powers of the existing regulators and conclude that enforcement of existing legislation and self-regulation cannot be relied on to meet public expectations of greater accountability.
  7. [ . . . ]