ICGN Investor Viewpoint 

Artificial intelligence: An engagement guide 

1. Introduction 

Artificial intelligence (AI) presents both extraordinary opportunities and complexities for today’s companies. The OECD defines an AI system as “a machine-based system that, for explicit or implicit objectives, infers, from the input it receives, how to generate outputs such as predictions, content, recommendations, or decisions that can influence physical or virtual environments”.¹ An increasing number of companies are using AI to transform existing business models or create new ones, generate greater efficiencies, and enhance strategic decision-making, all of which are critical for their competitiveness. However, AI also poses risks and challenges that company boards and management teams must be able to understand and address.  

Investors expect companies to effectively navigate AI-related challenges whilst maximising the benefits of AI integration. Our Investor Viewpoint aims to encourage a constructive dialogue on this fast-evolving and increasingly important technology. We consulted the available sources and engaged with our members to provide a guide for investors and companies. This Viewpoint supports investors in assessing whether a company uses AI in a safe, ethical, and sustainable manner, leading to a series of questions for use in investors-investee dialogue. Also, by proactively using this guide, boards can anticipate investors’ areas of interest and concern, and better assess the robustness of their AI oversight. 

2. Investors’ expectations of companies 

The rapid development and use of AI necessitates comprehensive AI governance and risk management processes.²  Board oversight, responsible practices, risk management and accountability to shareholders and stakeholders, and transparency and explainability, are cornerstones for responsible AI.³ 

Board oversight 

The board of directors (board) is accountable for overseeing a company’s responsible development and use of AI.⁴ As part of its fiduciary duty to preserve and enhance long-term  value, the board should ensure that the company management balances the competitive deployment of new technology against potential risks – including risks to people and society. 

Boards should be able to explain to investors the extent to which the company approaches AI as a risk or as an opportunity, and its short, medium, and long-term plans for integrating AI as part of its business model. 

Boards should ensure that they are properly equipped to oversee AI related risks and opportunities. Knowledge on AI can come from different channels, including trainings, advisory bodies, engagement with external experts, and continuous awareness programmes. Surveys suggest that many boards overestimate or lack AI knowledge.⁵ If board members do not have sufficient AI expertise or lack access to advice, they may be ineffective in identifying and assessing material AI related risks and opportunities.⁶ 

Responsible AI practices 

Companies should implement AI in a way that preserves trust in the company and prevents, as far as reasonably possible, economic, human, social, and environmental harm. Companies should implement AI governance and due diligence procedures proportionate to the potential impacts of their AI activities. Companies developing and training AI face distinct risks from those using third-party AI solutions, and bear responsibility for ensuring that their system programming promotes safe and ethical outcomes. Companies should envisage articulating their approach to AI in an overarching statement or a set of principles, and should embed responsible AI in existing policies, such as their Code of Conduct, Information Security, Data Ethics, Data Privacy and Vendor Assessment Policies. 

AI’s impact on the workforce is a complex and evolving topic: AI raises major concerns about job displacement, changes in job roles, and the need for new skills. Whilst research suggests that AI could expose millions of full-time jobs to automation,⁷ a study found that fewer than one in three CEOs had assessed the potential impact of generative AI (GenAI) on their workforces. GenAI refers to deep-learning models that can generate high-quality text, images, and other content based on the data they were trained on.⁸ Awareness of the potential societal impact of mass automation and understanding the effects of the company’s use of AI on its workforce is critical and should be discussed at board level.⁹ Companies should consider upskilling or reskilling efforts to address evolving talent needs.¹⁰  

Furthermore, companies should define the scope of AI applications used in human capital management. This should reflect whether the company believes the technology could enable fairer, more efficient, transparent, and inclusive practices, how the technology is expected to deliver on this, and what might happen if it fails.¹¹ Companies should ensure that employees are trained to use AI in a responsible manner, with a good understanding of the technology and its potential impacts. Companies must address potential skewing of AI benefits distribution amongst the workforce and mitigate accessibility barriers for employees with disabilities.¹² 

Regarding environmental risks, research shows that AI could support natural capital management through positive innovations. Identifying pollution, and mapping and measuring deforestation and the melting of icebergs are established examples.¹³ However, AI uses more energy than other forms of computing, raising concerns about its own ecological footprint.¹⁴ Training a single model uses more electricity than 100 homes consume in a year, and semiconductor production (which forms part of AI’s underlying computational hardware) is highly polluting.¹⁵ These considerations must be included in companies’ decision-making. 

Risk management 

Boards should ensure that companies have strong risk management processes to identify, assess and mitigate financially material AI-related risks, as well as potential adverse impacts on society and the environment.  

AI related risks include, among others: 

• Unwanted bias, when automated systems relying on biased data or design produce discriminatory outcomes, perpetuates inequalities in decision-making. Some companies have faced legal action after using AI systems allegedly reinforcing discriminatory outcomes.¹⁶  

• “Hallucinations”, referring to when AI generates false information.¹⁷ 

• AI systems trained on inaccurate, outdated, or otherwise not fit for purpose data.¹⁸ 

• Spread of mis/dis-information or harmful content through AI generated content.   

• Failure to evaluate risks of third-party AI. Research suggests that more than half of all AI failures come from third-party tools, which most companies rely on.¹⁹ 

• Intellectual property (IP) infringement.²⁰  

• Data security breaches, including hacking or privacy violations. 

• Technical malfunctioning, causing autonomously operated machines to endanger 
  human life, for instance. 

To mitigate risks, it is advisable to utilise AI under human supervision and intervention. Also, initiating an AI use-case pilot together with sector, market, and product specialists before integrating the system across the business can help identify areas of success and concern.²¹ 

To identify and address salient human rights issues or adverse impact on the environment, boards should ensure that management conducts impact assessments, audits, and due diligence.²² Understanding the implications of the use of AI for human rights to privacy and data protection, equality and non-discrimination, and for social cohesion, for instance, are fundamental elements of AI governance.²³  

More generally, companies should conduct risk-based due diligence throughout their value chain, in line with the OECD Guidelines for Multinational Enterprises, and this should include AI-related considerations. This entails: (1) embedding responsible business conduct into policies and management systems, (2) identifying, and (3) ceasing, preventing, or mitigating potential or actual adverse impacts, (4) tracking results, (5) communicating, (6) providing grievance mechanisms and remediation when appropriate.²⁴ 

Transparency and explainability 

Transparency and explainability helps build trust and ensure accountability to shareholders, stakeholders, and society at large. Company management should be able to explain to their boards how the AI systems they develop or use have been designed, trained, tested and scaled, and how they align with human values and intent.²⁵ Companies developing and training AI must be transparent about what data the model has been trained on.  

Furthermore, all companies should be transparent about how the AI systems they deploy collect, use, and store personal data.²⁶ According to emerging best practices, company management should ensure that stakeholders, such as customers and employees, have consented to their data being used by AI. Moreover, all stakeholders should be made aware of their interactions with AI systems and companies should be transparent about any content that is AI-generated.²⁷ Finally, investors expect timely disclosures of any material AI-related controversy.  

Regulatory compliance 

Whilst most countries do not, at the time of writing, have AI-specific regulation, there are frameworks that companies should follow (see Annex 1). Boards should ensure that company management implement existing regulation and relevant standards on responsible AI, such as OECD’s AI Principles, UNESCO’s Recommendation on the Ethics of AI, or ISO/IEC 421001:2023, which specifies requirements for establishing, implementing, and improving AI management systems in an organisation.²⁸ As AI regulation is an evolving policy area, boards and management should stay up to date with latest developments.  
 
3. Stewardship dialogue 

ICGN encourages shareholders to engage in a constructive dialogue with investee companies, with the objective of creating long-term value on behalf of beneficiaries or clients. When engaging with company boards and management teams on AI-related matters, investors can consider the following points, which boards can also use to better assess the robustness of their AI oversight. 

1. Has AI been considered in the development of the company’s strategy? Does the company (or is it planning to) develop or use AI and, if so, how?  

2. How does the board ensure that it has sufficient knowledge and understanding of AI, if deemed relevant for the company? 

3. Did the company publicly articulate its approach to responsible AI? Is responsible AI embedded in relevant company policies (e.g. Code of Conduct, Data privacy)?  

4. Which risk management processes were established to identify material AI-related risks and mitigate these? What are the key AI related risks for the company, and how are they being mitigated? Who would be held responsible for AI controversies? 

5. Have any biases and privacy issues been identified?  

6. Has the board discussed how the AI systems the company uses or develops have been designed, trained, and tested?  

7. How does the company assess the implications of its use of AI on the workforce? 

8. Is management planning to reskill or upskill employees affected by automation, and, if so, how will success be measured?  

9. How does management conduct risk-based due diligence to identify, and prevent or mitigate adverse impacts of its use of AI on society and the environment?  

10. How regularly does the board engage with its stakeholders on AI, including employees? Did the company establish a grievance mechanism for AI-related matters? 
     

4. Conclusion  

AI holds profound promise for boosting business efficiencies, productivity, and deriving intelligence, and companies failing to deploy AI may be exposed to a loss of competitiveness. However, there are material risks associated with the use of AI, and investors expect companies to use AI responsibly.  

Boards should ensure the implementation of robust AI governance processes encompassing oversight and accountability, responsible AI practices, transparency and explainability, robust risk management, and regulatory compliance to support safe, ethical, and sustainable development and use of AI.  

With the fast pace of technological change, and AI being used by an increasing number of companies across many economic sectors, best practices for responsible AI – in terms of governance, conduct, and reporting – will continue to evolve. It is in this spirit that ICGN encourages a constructive dialogue between investors, companies, policymakers, and standard-setters on this important topic. 

About this Viewpoint  

ICGN Viewpoints provide opinion on emerging corporate governance issues and are intended to generate debate, whilst not defining a formal ICGN position on the subject. This Viewpoint is an update to ICGN’s Viewpoint on AI and Board Effectiveness (2020). 

ICGN is grateful for the support of the ICGN Human Capital Committee, ICGN Global Governance Committee, the 695^th Lord Mayor’s Ethical AI Initiative, Christine Chow (ICGN board member), Nicholas Beale (Sciteb), Elisa Cencig and Emily Nathan (NBIM), and Aldo Bonati (Etica) who contributed to this Viewpoint. 

We encourage dialogue by contacting Severine Neervoort, Global Policy Director, severine.neervoort@icgn.org, or Wendela Rang, Policy Executive, wendela.rang@icgn.org.Annex 1: Standards and regulation on AI, as of March 2024 

Global: OECD AI Principles set a global framework for responsible AI. The Asilomar AI Principles outlines 23 guidelines for AI research and development. ISO/IEC 421001:2023 is designed for entities providing or utilising AI-based products or services. IEEE SA – IEEE 7000-2021 Standard Model Process for Addressing Ethical Concerns during System Design sets out processes for ethical values consideration in AI development. BSA The Software Alliance compares progress on international AI frameworks.  

Europe: The EU institutions reached a provisional agreement on the EU Artificial Intelligence Act in December 2023. In the UK, there is no national legislation. The UK Government’s ‘Guidance on AI and Data Protection’ aims to help organisations adopt technologies whilst protecting people and vulnerable groups.  The Government’s ‘White Paper on AI Regulation: A Pro-Innovation Approach’ explains how the Government aims to support innovation whilst providing a risk management framework.  

North America: In the U.S, there are various legislative AI activities at federal and state level. State consumer protection laws are widely viewed as applying to AI. New York City’s Local Law 144 regulates automated employment decision tools. The U.S Executive Order on AI sets out standards for AI safety and security. The National Institute of Standards and Technology (NIST) AI Risk Management framework offers best-practice guidelines. Canada’s Generative AI Code of Conduct sets out voluntary guidelines. The Canadian government proposed a law to regulate AI in September 2023.  

Asia and Australia: Only China has national AI regulation. Many countries have guidelines and frameworks, such as Japan’s ‘Social Principles of Human-centric AI’ and Singapore’s ‘Model Framework’. China introduced a law to regulate GenAI, mandating that GenAI should not be discriminatory on the basis of race or sex, among other. In Hong Kong, the Privacy (Data) Protection Ordinance (PDP) applies to AI. Explore AI laws in Indonesia, Malaysia, Singapore, Thailand, and Vietnam here. Australia is expected to regulate high-risk areas of AI following a recent consultation.  

Africa: Mauritius and Egypt have national AI strategies.¹⁸ The African Union (AU) Data Policy Framework guides the development of national data policies in AU member states.   

South, Central and Latin America: The Latin American AI Index, launched by the Economic Commission for Latin America and the Caribbean (ECLAC), analyses AI regulation in Argentina, Bolivia, Brazil, Chile, Colombia, Costa Rica, Ecuador, Mexico, Panama, Paraguay, Peru, Uruguay.  
 

Annex 2: Recommended reading 

• The AI Governance Alliance, ‘Briefing Paper Series’ (18 January 2024) 

• The Alan Turing Institute, ‘Data science, artificial intelligence, and the futures of work’ (October 2018) 

• Appel, Gill et al., “Generative AI has an intellectual property problem”, Harvard Business Review (07 April 2023) 

• Chatham House, ‘AI governance and human rights: Resetting the relationship’ (January 2023) 

• Chris Gillam, “Will AI change everything?”, CPP Investments Insights Institute, (9 June 2023) 

• Chow, Christine et al., “Future of Work: Investors’ Expectations on Ethical Artificial Intelligence in Human Capital Management” (2022) 

• Leslie, David, ‘A guide to AI ethics, including responsible design and implementation of AI systems in the public sector’, The Alan Turing Institute (2019) 

• Edelman, David and Vivek Sharma, “It’s Time for Boards to Take AI Seriously”, Harvard Business Review (02 November 2023) 

• Gregory, Holly J. “AI and the Role of the Board of Directors”, Harvard Law School Forum on Corporate Governance (07 October 2023) 

• Harvard Business Review, “AI and machine learning” (regularly updated) 

• Institute of Business Ethics, “Business Ethics and Ethical AI: Scrutinising AI with an ethical lens”  (26 October 2023) 

• Information Commissioner’s Office, “How do we ensure transparency in AI?” (15 March 2023) 

• MIT Sloan Review, “Third-party AI tools pose increasing risks for organization” (21 September 2023) 

• MIT Technology Review Insights, “Artificial Intelligence” (regularly updated) 

• NBIM, “Responsible artificial intelligence” (15 August 2023) 

• OECD, ‘Recommendation of the Council on Artificial Intelligence, OECD/LEGAL/0449 (2023) 

• OECD, ‘Artificial Intelligence and Responsible Business Conduct’ (n.d) 

• OECD, ‘OECD Guidelines for Multinational Enterprises on Responsible Business Conduct’, OECD Publishing (2023) 

• UNESCO, ‘Recommendation on the Ethics of Artificial Intelligence’ (2021) 

• U.S Government Accountability Office, “Workforce Automation: Insights into Skills and Training programs for Impacted Workers” (17 August 2022) 

• Will, Paris et al., ‘People versus machines: introducing the HIRE framework Artificial Intelligence Review’ (2023) 

• World Benchmarking Alliance, Digital Inclusion Benchmark (regularly updated) 

World Benchmarking Alliance, “Investor Statement on Ethical AI” (26 April 2022)