FAT Approaches Other Stakeholder Groups Can Implement

This section outlines approaches and benefits to promoting FAT around high-risk algorithmic systems that other stakeholders—such as investors, investor alliances, funders, and philanthropic organizations—can pursue.

Civil society groups, researchers, and advocates have found some success in pushing internet platforms to demonstrate transparency and accountability around their operations. But, they have thus far found less success when it comes to obtaining FAT around these companies’ algorithmic systems. Investors can play a critical role in this area.

Following the attack on two mosques in Christchurch, New Zealand, the country’s $41 billion sovereign wealth fund launched the first global investor coalition focused on engaging with social media companies to combat extremism and other prominent tech policy issues.¹ The coalition brought together over 100 investors with more than $13 trillion in assets to press companies to address issues related to harmful content.² This movement marked a scaling up in the role investors play when it comes to environmental, social, and governance (ESG) issues at technology companies. As social responsibility continues to play a growing role in business activity, investors are becoming increasingly concerned around how big tech handles issues such as data privacy, corporate surveillance, algorithmic harms, and content moderation. To this end, in 2019, the Sustainability Accounting Standards Board (SASB) launched a research project³ to explore whether establishing standardized metrics on content moderation of internet platforms was warranted.⁴ If advocates invest time in lobbying investors and standards setting organizations to pay greater attention to algorithmic harms, it could create a landscape ripe for investor pressure for greater FAT from internet platforms that use high-risk algorithmic systems. Standards-setting bodies could develop categorizations of algorithmic complexity, predictability, explainability, performance design, liability standards, best practices, and so on.⁵ These standards should be developed in conjunction with civil society groups, academics, internet platforms, and other stakeholders, and should address how internet platforms develop and deploy algorithmic systems.

Investors seeking to push internet platforms for more FAT around their algorithmic systems, need to collaborate to develop clear standards related to algorithmic systems that they want companies to commit to meeting. In addition, the investor community must produce well-established accountability mechanisms that enable them to gauge whether potential and current portfolio investments are meeting their standards around algorithmic harms. Both the standards and accountability mechanisms should be developed based on feedback and consultation with a broad range of stakeholders, including civil society groups and researchers. If widely adopted, this practice could compel emerging internet platforms to adopt responsible FAT policies around algorithmic systems, or risk losing access to capital they need to survive. If this practice is not too burdensome and becomes common among smaller companies, it could then put pressure on bigger companies that do not rely as heavily on investment sources to adopt responsible practices as well.⁶

The impact investing community has begun developing standards and accountability mechanisms that allow investors to ensure their investments are generating positive social and environmental outcomes. These mechanisms also enable investors to track whether businesses they invested in are living up to their ESG commitments.⁷ Investors and investor alliances have also successfully pushed for change in corporate policies and practices related to workplace discrimination as well as climate change and resource management issues such as deforestation and water management.⁸

In addition, investors, funders, and philanthropic organizations can help establish critical structures and functions related to FAT promotion, especially when it comes to the use of high-risk algorithmic systems. For example, investors, funders, and philanthropic organizations could direct funds toward stakeholders working to increase technical expertise in auditing entities and the government, and to establish robust independent auditing mechanisms for high-risk systems⁹

These entities could also devote greater funding toward stakeholders working to train journalists, social science researchers, and civil society organizations to become more algorithmically literate, and develop educational materials to educate policymakers and the public about the harms caused by high-risk algorithmic systems and potential mitigation strategies.¹⁰ These kinds of targeted investment or funding arrangements can help to establish a more comprehensive landscape of actors working on promoting FAT around high-risk algorithmic systems and help push the FAT agenda forward.