Developing a comprehensive approach for algorithmic accountability to protect human rights.
Modern algorithms are replacing human decision making. Algorithms conduct sophisticated predictive analytics and execute complex tasks beyond human capability and speed. Their application is used to automate many functions traditionally carried out by humans and has expanded to key areas of decision-making. This includes algorithmic assessments in applications such as sentencing decisions, credit scoring, recruitment, and social security. The use of algorithmic systems to make or support decisions is becoming increasingly central to many areas of public and private life. This can affect all our human rights, from civil, cultural, economic, political, to social rights.
The pace of technological innovation is faster than the formulation, application and enforcement of governance and regulation of algorithms in decision-making. Some commentators have suggested that governance and regulatory mechanisms are anti-innovation and that it is too late or too difficult to manage this area of technological innovation. This exceptionalises new technologies such as algorithmic and artificial intelligence systems. It is not a valid argument for failing to govern the development and use of algorithmic systems. Human rights are universally applicable. New technologies can impact on human rights like any other sector and the potential benefits for such systems do not discount the need for ensuring human rights are respected and protected. International human rights law applies to the use of new technologies just as it applies in any other area of life.
The human rights-based approach to algorithmic accountability
States and businesses engaged in any part of the algorithmic life cycle, from the design, development and deployment to the supply of algorithmic systems, should embed a human rights-based approach.
International human rights law provides a means to define and assess harm, and provides a deeper and fuller means of analysing the overall effect of the use of algorithms. The specific obligations on States and expectations on businesses to prevent and protect human rights includes prescription of the mechanisms and processes required for implementation. The international human rights law framework can map on to the algorithmic life cycle and offers a holistic approach for accountability.
Existing mechanisms for algorithmic accountability such as data protection, impact assessments and compliance checks may have some relevance for protecting human rights and preventing violations. The international human rights law framework complements these frameworks and contributes to a more comprehensive approach for algorithmic accountability, incorporating robust safeguards and assessing the full scope of impact.
The next steps for our research are to operationalise this framework in practical guidance for states and businesses.
Ensuring that Human Rights are central to AI regulation and governanceRead more
A key part of HRBDT’s work is to show how human rights are both affected by the design, development and…
- Laura Carter
Google’s pay audit and the meaning of ‘equality’Read more
This blog was originally published on the Human Rights Centre Blog. In March of this year, Google found that they were…
- Lorna McGregor
The need for clear governance frameworks on predictive algorithms in military settingsRead more
This post originally appeared on the Humanitarian Law and Policy Blog. Editor’s note: In this post, as part of the AI blog…
- Latest #podcast from MIT Technology Review: 'What happens when an #algorithm gets it wrong?' Hear @PeteFussey dis… https://twitter.com/i/web/status/1293921040976900097
- 11/11 Read the full Met Police trial report - http://bit.ly/LFRreport Follow @HRBDTNews for updates.… https://twitter.com/i/web/status/1293135196787748864
- 10/11 "What is abundantly clear is that all police forces should pay greater attention to human rights law consider… https://twitter.com/i/web/status/1293135195336474624
- 9/11 (contd.) "However, many issues remain to be addressed, including the broader societal impact of facial recogni… https://twitter.com/i/web/status/1293135193721786368
- 8/11 (contd.) "are consistent with many of our own conclusions regarding the Metropolitan Police Service. This is a… https://twitter.com/i/web/status/1293135192304029699
- 7/11 (contd.) "...in terms of how individuals develop and interact and how challenges to, or protests against, gove… https://twitter.com/i/web/status/1293135190269857792