Despite all its promises, Artificial Intelligence (AI) presents risks to human rights online and offline, and growing ethical challenges. To benefit humanity, significant measures are needed, combining positive encouragements and stricter rules.  

The AI Revolution

The Fourth Industrial Revolution is driven by a convergence of AI and other disruptive technologies. AI is a moving target and an umbrella term referring to what machines have not achieved yet. Given set of objectives, AI makes recommendations, predictions or decisions. Like electricity, it is a general-purpose technology, impacting multiple industries and areas of global value chains. With cloud computing, exponential growth in data and algorithmic progress, it affects how we plan, produce, distribute and consume. PwC estimated that by 2030, AI will add up to $16 tn. to the global economy. AI pervades our everyday world: Machine Learning for GPS and Netflix; Deep learning suggests email replies and unlocks smartphones with facial recognition. AI outperforms the best game players and cracks 50-year old biological challenges. AI also contributes to the UN Sustainable Development Goals

AI is one of the most profound things we’re working on as humanity. It’s more profound than fire or electricity.

Sundar Pichai, Alphabet’s CEO

The Promise of AI

AI helps solve complex problems like fast tracking vaccine development and delivery. Through better predictions, recommendations or decisions, the promise is to enhance efficiency and sustainability and build back better. From performing repetitive or dangerous tasks to predicting climate events, AI’s adoption varies greatly across sectors. It is deeper where patterns are revealed within datasets and efficient models built to enhance decision-making.

• Acceleration of scientific discoveries
• Better disease prevention, diagnosis and treatment
• Cleaner and safer transportation
• Sustainable management of environmental resources
• Optimised use of pesticides and fertilizers
• Real-time cyber and digital security
• Sustainable supply chains and public services

AI is ubiquitous in our society. As with any new technology, opportunities are moderated by concerns about ethical, human rights challenges. AI can save lives and mitigate climate risks, but if unchecked, it can deepen inequalities.

How do we maximise its potential for good while addressing ethical and human rights repercussions? Borrowing from Greek mythology, if AI is a gift from God, how do we avoid the curse when opening Pandora’s box.

AI can be a great opportunity to accelerate the achievement of sustainable development goals. But any technological revolution leads to new imbalances that we must anticipate.

Audrey Azoulay, UNESCO Director-General

AI Gaps

There is no universal definition of ethics. Divergent interpretations are based on culture and perspective. AI ethical standards and guidelines abound and remain voluntary. Human rights are universally recognised and enshrined in legally binding treaties (ICCPR, ECHR). For AI to respect ethics and human rights, several gaps should be addressed:

• Human-centric, ethical and trustworthy: are we putting humans first when planning, designing, building, training, deploying and governing AI systems?
• Bias, discrimination and fairness: are we propagating bias with datasets used to train algorithms? How transparent and explainable are the decisions?
• Surveillance, data rights and privacy: is AI monitoring or profiling behaviours without accountability and consent? Arerights to be forgotten or remain anonymous preserved?
• Manipulation, freedoms and democracy: are fake news discrediting individuals and organisations or influencing elections? Is content moderation infringing on freedom of speech?
• Life, dignity, peace, and security: should we delegate policing and justice to algorithms? Is it ethical to entrust life and deaths decisions with autonomous drones?
• Work, automation and digital welfare: is AI massively replacing or displacing jobs? Should we outsource social protection to algorithms?

Ethics and human rights are needed to build effective safeguards. Common rails promote scalable innovation across sectors while guardrails prevent or sanction misuse and abuse by Big Tech.

Unpacking Pandora’s Box

Given the interdependence and complexity of the issues, multiple approaches should be combined. AI is not a topic for technologists that we sprinkle with some ethical and human rights magic powder. Let’s integrate them from the beginning, so regulation and self-regulation are used in synergy without stifling innovation.

Awareness & Training

As a sociotechnicalsystem, AI depends on goals, datasets and contexts in which it is deployed. Impacts (positive or negative) are a reflection of the designers and operators values. Diversity in AI teams matters. AI community members need training on inclusion, ethics and human rights, so they are embedded in models and datasets throughout AI’s life cycle. For policymakers, AI literacy is required to communicate effectively with developers. Making the right policy calls will benefit their constituents.

Self-regulation

AI developers are encouraged to adhere to ethical principles or codes of conduct set by industry leaders (MSFT; Google; Deepmind) and standard setting organisations (IEEE, ISO). This will reflect their commitment to consumers and governments, and may lead to a competitive advantage. Inspired by Isaac Asimov laws of robotics, China recently unveiled its Governance Principles for AI development and UNESCO adopted an agreement on AI ethics.

Policy-making

AI characteristics make it complex to regulate. Policy-makers need to balance protecting society and fostering innovation. Many call for immediate legislation to regulate Big Tech, akin to unaccountable public utilities that do enormous good but also harm. In April 2021, the European Commission proposed a regulation banning or restricting applications based on assigned risk levels. This could have the same global impact as the GDPR. China’s Cyberspace Administration released its draft proposal to regulate content recommendation systems. If confirmed, it would expand government’s control over data flows and speech.

AI is neither a curse nor a gift from God. We can build institutions to shape AI in a safe, ethical and responsible way so our human rights are upheld. It takes concerted efforts by developers, policymakers and civil society, working collaboratively to realise the promise of a human-centric AI.

Mouloud Khelif, INP’21
Consultant – International Strategy and Sustainability
Executive Master INP (2021) – Master of Advanced Studies (2022)
mouloud.khelif@graduateinstitute.ch
Twitter: @Mouloudkhelif
Linkedin

---

Azhar A (2019) Tony Blair: Governing in the age of AI [podcast]. Exponential View. October. Available at https://open.spotify.com/episode/3EF7y7O4JNdaM4RGeOlDwy [accessed 25 April 2021].

Bailey K (2016) Reframing the ‘AI effect’. Medium blog, 27 October. Available at https://medium.com/@katherinebailey/reframing-the-ai-effect-c445f87ea98b [accessed 16 February 2021].

Brenneis E (2019) Dealing with digital disruption: How businesses can gain a competitive advantage. Tech Radar, 21 March. Available at https://www.techradar.com/news/dealing-with-digital-disruption-how-businesses-can-gain-a-competitive-advantage [accessed 28 April 2021].

Carnegie Mellon University CMU (2019) Carnegie Mellon and Facebook AI Beats Professionals in Six-Player Poker, 11 July. Available at https://www.cmu.edu/news/stories/archives/2019/july/cmu-facebook-ai-beats-poker-pros.html [accessed 25 April 2021].

Council of Europe Ad Hoc Committed on Artificial Intelligence CAHAI (2020) Feasibility Study, 17 December. Available at https://rm.coe.int/cahai-2020-23-final-eng-feasibility-study-/1680a0c6da [accessed 25 April 2021].

Council of Europe CoE (2019) Unboxing artificial intelligence: 10 steps to protect human rights, Commissioner for Human Rights Recommendation. 14 May. Available at https://www.coe.int/en/web/commissioner/-/unboxing-artificial-intelligence-10-steps-to-protect-human-rights [accessed 25 April 2021].

European Commission EC (2021) Europe Fit for the Digital Age: Commission proposes new rules and actions for excellence and trust in Artificial Intelligence, 21 April. Available at https://ec.europa.eu/commission/presscorner/detail/en/ip_21_1682 [accessed 23 April 2021].

Gill A (2020) Imagining the AI future, IISS Survival Editor’s Blog, 2 January. Available at https://www.iiss.org/blogs/survival-blog/2019/12/imagining-the-ai-future [accessed 26 April 2021].

Jobin A (2020) Ethics guidelines galore for AI – so now what? ETH Zurich, 17 January. Available at https://ethz.ch/en/news-and-events/eth-news/news/2020/01/ethics-guidelines-galore-for-ai.html [accessed 28 April 2021].

Kahn J (2021) Europe proposes strict AI regulation likely to have an impact around the world, Fortune, 21 April. Available at https://fortune.com/2021/04/21/europe-artificial-intelligence-regulation-global-impact-google-facebook-ibm/ [accessed 29 April 2021].

Kunz L (2021) Build Back Better with Responsible AI, KI – Künstliche Intelligenz 35, pp. 1–3. Available at https://doi.org/10.1007/s13218-021-00707-9 [accessed 28 April 2021].

Lynch S (2017) Andrew Ng: Why AI Is the New Electricity. Stanford Business, March 11. Available at https://www.gsb.stanford.edu/insights/andrew-ng-why-ai-new-electricity [accessed 25 April 2021].

Mayer H et al. (2020) AI puts Moderna within striking distance of beating COVID-19. HBS Digital, 24 November. Available at https://digital.hbs.edu/artificial-intelligence-machine-learning/ai-puts-moderna-within-striking-distance-of-beating-covid-19/ [accessed 25 April 2021].

Mayor A (2018) What Pandora’s Box tells us about AI. World Economic Forum (WEF), 19 October. Available at https://www.weforum.org/agenda/2018/10/an-ai-wake-up-call-from-ancient-greece/ [accessed 29 April 2021].

PwC (2017) Sizing the prize. PwC Global AI Study: Exploiting the AI Revolution. Available at https://www.pwc.com/gx/en/issues/data-and-analytics/publications/artificial-intelligence-study.html [accessed 27 April 2021].

Sample I (2020) DeepMind AI cracks 50-year-old problem of protein folding. The Guardian, 30 November. Available at https://www.theguardian.com/technology/2020/nov/30/deepmind-ai-cracks-50-year-old-problem-of-biology-research [accessed 27 April 2021].

Teleanu S (2019) Artificial intelligence: policy implications, applications, and developments. Digital Watch Observatory. Available at https://dig.watch/topics/artificial-intelligence/  [accessed 26 April 2021].

The Economist (2020), Technology Quarterly. An understanding of AI’s limitations is starting to sink in, 13 June. Available at https://www.economist.com/technology-quarterly/2020/06/11/an-understanding-of-ais-limitations-is-starting-to-sink-in  [accessed 25 April 2021].

UNESCO (2021) UNESCO member states adopt the first ever global agreement on the Ethics of Artificial Intelligence, 25 November. Available at https://en.unesco.org/news/unesco-member-states-adopt-first-ever-global-agreement-ethics-artificial-intelligence [accessed 30 November 2021]

Vinuesa et al. (2020) The role of artificial intelligence in achieving the Sustainable Development Goals, Nature Communication 11, 233, 13 January. Available at https://doi.org/10.1038/s41467-019-14108-y [accessed 28 April 2021].

World Economic Forum [WEF] (no date) Shaping the Future of Technology Governance: AI and Machine Learning. Available at https://www.weforum.org/platforms/shaping-the-future-of-technology-governance-artificial-intelligence-and-machine-learning [accessed 26 April 2021].