Aligning Artificial Intelligence (AI) with Human Rights: Key Principles and Themes

Author:

Erica Giraldo Toro - Lawyer & Managing Director of iGlobal.Lawyer

Foundational Source of the article

This article is grounded on the work of the known report, carried out by Berkman Klein Center for Internet & Society, named: “Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-based Approaches to Principles for AI”, led by Jessica Fjeld, the assistant director of the Harvard Law School Cyberlaw Clinic, to which many esteemed stakeholders such as the UN High-Level Committee on programmes and the OECD’s Expert Group on AI, have contributed.

Abstract

Artificial Intelligence (AI) technology is currently in the spotlight, as we are realizing that it has the potential to be a game-changing protagonist in both negative and positive ways.

The normative path that we take is crucial to creating an ethical and Human Rights-respecting AI.

At present, there is still a significant degree of uncertainty when it comes to developing these new "intelligent" systems with a human-centered approach because the word “bias” seems to be a recurring problem in AI.. however, there is a good news: things could change if we proceed with implementing some core principles that prioritize the human being in any single action taken in this direction.

We drew inspiration from the source mentioned at the beginning of this reflection, because we believe it provides a solid foundation for responding to an essential question: 

what general principles has the international community agreed upon to ensure the alignment of Human Rights with AI development and deployment?

AI & Human Rights: Core Concepts

The source outlines 47 principles, categorized into eight key themes, gathered from 36 documents drafted by leading authorities and organizations across different continents, concerning the way AI should be developed, deployed and governed.

This approach embodies the principle of Universality of Human Rights, a very well-known concept in the field of International Human Rights Law that now - in the 21st century - must be extended to AI as well.

This article aims to provide a broad overview of these eight key themes with their related principles, serving as the very starting point for our focus in the next months through a step-by-step analysis.

1. Privacy

Principles under this theme: Consent; Control over the Use of Data; Ability to Restrict Processing; Right to Rectification; Right to Erasure; Privacy by Design; Recommends Data Protection Laws; (Other/General) Privacy.

The impact of technology on our privacy is substantial, vital and of paramount importance.

The General Data Protection Regulation (GDPR) of the European Union (EU), the most stringent privacy and security law globally, establishes two categories of compliance: data protection & data privacy, underscoring the necessity of protecting personal data as a fundamental milestone of any democratic society.

Consequently, discussing AI’s impact on our privacy and the significance of binding laws at international, national, or regional levels remains a top priority in the general debate.

2. Accountability

Principles under this theme: Verifiability and Replicability; Impact Assessments; Environmental Responsibility; Evaluation and Auditing Requirement; Creation of a Monitoring Body; Ability to Appeal; Remedy for Automated Decisions; Liability and Legal Responsibility; Recommends Adoption of New Regulations; Accountability Per Se.

AI is often associated with human intelligence, but the reality is that human oversight will always remain essential.

The question that now arises is: who is accountable for the decisions made by autonomous AI?

The vast potential of technology demands accountability, as it impacts the natural world and transcends geographical boundaries.

3. Safety and Security

Principles under this theme: Safety; Security; Security by Design; Predictability.

AI must ensure the integrity, confidentiality and privacy of personal data, otherwise we - users - are going to lose trust in AI technology and we can’t let this happen.

The primary goal of any AI creation should be to protect human beings and their personal integrity, identity and dignity.

4. Transparency and Explainability

Principles under this theme: Transparency; Explainability; Open Source Data and Algorithms; Open government Procurement; Right to Information; Notification when AI makes a decision about an individual; Notification when interacting with AI; Regular Reporting.

Numerous studies, including some from Harvard, point out the cruciality of providing competent authorities with access to source code because as AI increasingly permeates important fields for our personal development such as work and education, understanding AI processes and ensuring information transparency are crucial societal challenges that we have to tackle since the moment we put these new tools on the market.

5. Fairness and Non-Discrimination

Principles under this theme: Non-discrimination and the Prevention of Bias; Representative and High Quality Data; Fairness; Equality; Inclusiveness in Impact; Inclusiveness in Design.

AI should be accessible to all, regardless of culture, gender, sexual orientation, ethnic or social origin, language, religion or belief, political or any other opinion, profession; etc.

AI-based systems must be created ethically, also respecting the rights of marginalized groups which have always been more vulnerable to discrimination.

6. Human Control of Technology

Principles under this theme: Human Review of Automated Decisions; Ability to Opt out of Automated Decisions; Human Control of Technology.

Humans interacting with AI must maintain their agency and autonomy, it is therefore crucial that we retain control over our lives and decisions, in any stage and at any level.

7. Professional Responsibility

Principles under this theme: Accuracy; Responsible Design; Consideration of Long Term Effects; Multi-stakeholder Collaboration; Scientific Integrity.

AI systems should be built on values and professional practices, embodying professional responsibility that we naturally have but that, machines, don't.

8. Promotion of Human Values

Principles under this theme: Human values and Human Flourishing; Access to Technology; Leveraged to Benefit Society.

AI should adhere to ethical standards, focusing on enhancing individuals rather than replacing them and, at the same time, AI should not also contribute to stress, anxiety or other adverse effects but, on the contrary, AI must respect human dignity and protect individual rights and freedoms.

Conclusion

We recognize that the Fundamental Rights and freedoms enshrined in documents like the Universal Declaration of Human Rights (UDHR) must extend to the all AI-related matters because they actually provide us a roadmap for ensuring AI is developed and deployed in a way that enhances rather than undermines human agency, dignity and wellbeing.

However, translating these principles into practice remains a significant challenge as it is not yet completely clear, either. What's clear, however, is that Ethics must apply to every stage of this revolution.

Embedding Human Rights and Ethics into the entire AI lifecycle, from data sourcing to model deployment, requires concerted efforts by developers, companies, policymakers and civil society, thus, robust governance frameworks, impact assessments, and multi-stakeholder collaboration and sustained commitment will be an essential element to ensure AI truly becomes a force for good in the 21st century.

Citation

Fjeld, Jessica and Achten, Nele and Hilligoss, Hannah and Nagy, Adam and Srikumar, Madhulika, Principled Artificial Intelligence: Mapping Consensus in Ethical and Rights-Based Approaches to Principles for Al (January 15, 2020). Berkman Klein Center Research Publication No. 2020-1, Available at SSRN: 

https://ssrn.com/abstract=3518482