The Stories We Tell About AI – Stanford Social Innovation Review

This post was originally published on this site.

(Photo by iStock/imaginima)

Narratives are an essential prerequisite to social change. The stories we see in the media and in content shared, both online and off, influence how society perceives, interacts with, and ultimately governs an issue area. Stories matter. We saw this in 2023, when the Hollywood strikes drew attention to AI in the workplace on an unprecedented scale. The narrative surrounding the strikes opened a discussion on the potential impact of AI on creative jobs, contributing to a win for workers. This was an important milestone.

However, for many industries, the stories being told about AI, work, and livelihoods serve to increase technology companies’ power at the expense of workers’ rights. These stories are championed uncritically by the media, are present in nascent industrial policies, and are reinforced by a majority of business leaders.

The figure below shows how many of the stories on AI we are accustomed to seeing in the media can be clustered into four broad categories. The first category contains stories related to AI’s impact on productivity. For example, a recent headline from The Times of India reads “Significant percentage of organisations see AI as key to competitive advantage: Report.” The second category relates to stories about AI’s complexity; as an example, a headline from Forbes reads “Artificial Intelligence: It’s Complicated And Unsettling, But Inevitable.” The third category contains stories related to the risks associated with AI technologies, like an article from the UK newspaper The Telegraph entitled “ChatGPT creator warns of the ‘existential risk’ from AI within a decade.” The fourth and final category pertains to stories about AI regulation and its potential impact on innovation. For example, a headline from Tech Policy Press reads “Why Hasty AI Regulation Could Hurt Africa.”

These stories, and the narratives they uphold, are problematic for workers everywhere.

FIGURE 1: Examples of Media Stories that Contribute to Dominant AI Narratives

Click to enlarge

Over the last year, the AI team at Purpose, a social impact creative agency and consultancy, has conducted detailed research to identify five priority areas to center workers in AI policy discussions and public discourse. The aim of such a worker-centered discourse is to ensure high-quality and secure livelihoods for all working people. Central to our approach has been a focus on shifting power dynamics and imagining the world of work in 2030 if existing AI narratives are not challenged.

In a previous SSIR article, Nobel Laureates Daron Acemoglu and Simon Johnson offer practical solutions for policy makers, business leaders, and organized labor movements to ensure AI works for workers. These proposed solutions, along with ample others being advanced in the United States and beyond, are imperative.

However, to create an enabling environment for systemic changes like these to take place, assumptions and beliefs need to be shifted about what is possible by showcasing alternative realities. This requires changing the stories we tell about AI. To ensure a more equitable future, workers must be at the center of narratives around AI and work.

Dominant AI Narratives Marginalize Workers

The Purpose AI team conducted over 50 interviews with technology, labor, and narrative experts, building on a review of the literature on AI narratives. We also convened a series of roundtable discussions and a stakeholder workshop that brought together thought leaders from the organized labor movement, multilateral organizations, and philanthropic foundations. We employed the Delphi method, iteratively refining the questions we asked, helping us to narrow down on prevalent AI narratives and their implications for workers.

From this research, we identified four overarching narratives that currently dominate the debate on AI (see FIGURE 2). Each narrative works in a different way to marginalize workers, and diminish the agency they have over decisions regarding the design, development, and deployment of AI at work.

FIGURE 2: Dominant AI Narratives, Supporting Assumptions and Implications for Workers

Click to enlarge

1. AI Will Make You More Productive and Efficient

The first narrative holds that by implementing AI in their operations, businesses will find greater efficiencies and scale. Business leaders, therefore, feel they have to move quickly to implement AI to remain competitive, thereby circumventing worker engagement in the process. This harms workers who have to use, interact with, or be surveilled by a technology designed by developers who do not fully understand the context of its deployment.

2. AI Is Complex

The second common narrative holds that AI technologies are a “black box,” that lay people can never understand how they work, and so the design and deployment of AI should be left to the “experts,” i.e. tech companies. The implication for workers is a disincentivization to engage with how new technologies shape their livelihoods.

3. AI Presents Existential Risk

The third dominant narrative focuses public attention on the potential existential risks of AI at some unknown point in the future. This focus prioritizes future risks over current ones. Many discussions on AI and work focus attention on a risk to the quantity of jobs in the future (i.e. job displacement) rather than on the quality of jobs now, thereby deprioritizing current harm to workers.

4. Regulation Stifles Innovation

The final dominant narrative furthers the idea that regulation is merely red tape holding technology companies back from creating products that will benefit society. This ultimately leads to weaker—or nonexistent—regulation that does not sufficiently protect worker rights.

Where Will We Be in 2030?

A key question we addressed in our research was what the world would look like for workers in 2030, if nothing was done to change current narratives on AI. The picture that interviewees painted was not positive for workers’ rights and it consists of five broad themes.

First, they emphasized a further consolidation of power amongst technology giants, employers, and elites. They highlighted that, unchecked, AI is likely to exacerbate ethnic, disability, social class, and gender inequities, creating further marginalization in the workplace.

Second, negative impacts on the quality of work are also likely to increase. For instance, algorithmic management systems are already being used by some companies. As their use becomes more widespread, the shift from human to AI-driven management systems will raise concerns about fairness, transparency and worker autonomy.

Third, data privacy is an area of particular ethical concern, with experts flagging data collection, data use, and worker consent as crucial challenges that will need to be addressed.

Fourth, if narratives remain unchanged, these impacts—and others—will continue to spiral as government regulation is sidelined under arguments of stifling innovation. This will allow technology companies to consistently undermine workers’ rights.

Fifth, labor union representatives highlighted that current workplace training on AI tends to focus on specific AI use cases—if at all—rather than preparing workers to interact with AI across multiple contexts over the course of their careers. If this limited education continues, it will minimize the touchpoints workers have with AI, reducing their ability to feel empowered and shape the way the technology is integrated as it continues to evolve.

To avoid this future, we need new narratives that center worker voices.

Five Priorities for Centering Workers in AI Narratives

To ensure a transition to AI that safeguards worker rights, narrows socioeconomic disparities, and enhances quality of life for working people, we must fundamentally shift the dominant narratives in public discourse. Purpose has been developing a Theory of Change (ToC) to guide this transformation, outlining a path to center workers in AI narratives and create a system that empowers workers to have a say in the design, development, and deployment of AI in their workplace. The following constitute five highlights where we believe change is most critical.

1. Developing an ecosystem of allies. Achieving this seismic shift in narratives requires engaging three key stakeholder groups, aligned with the International Labour Organization’s tripartite structure: workers, employers, and governments. To effectively transform narratives, we need an ecosystem of allies spanning different sectors of society, industries, and geographies. These allies will be instrumental in sharing success stories of worker-led AI integration and its positive outcomes—for workers and businesses alike.
2. A broader definition of a “worker.” Workers must be the leaders of this movement. For this to happen, we need to expand who is included in the category of “worker.” Given that AI touches all sectors, new language and messages are needed to move away from an outdated and narrow conception of the term (often those working blue collar jobs). Instead, we need an inclusive and expansive definition. One that enables a diverse coalition of workers—from lawyers and creatives to nurses and domestic workers—to come together and fight for secure futures and better-quality jobs.
3. Positioning workers as experts. Of equal importance is the need to position workers as experts in their respective fields. Current narratives diminish the expertise of those on the front lines, assuming technology companies can develop effective AI products without consulting the people who will be using them. Instead, we need to showcase labor expertise, uplift workers, and embed them in AI development processes. Some labor organizations have started this dialogue through formal agreements with tech companies. This is an essential step and the next step for all workers—unionized and not—should be to ensure this worker-tech dialogue becomes second nature for all companies.
4. Enabling AI narrative fluency. Many labor unions are highlighting the need for workers to be trained in the technical applications of AI. We agree this is essential. We also believe that workers should be trained in how to talk about AI, its impacts, and the narratives that surround it. The best messengers for sharing stories on AI and work are workers themselves. This is particularly important for building greater awareness amongst audiences—such as other worker groups—who might be skeptical of voices they do not perceive as aligned with their own interests, such as tech companies or CEOs. At Purpose, we call this the trusted messenger approach.
5. Highlighting success stories. In order for businesses—and ultimately society—to recognize the benefits of worker engagement in AI integration, we need to build a widespread normative understanding of its importance. To do this, labor unions and other worker representatives need to collect and share “win-win” stories that reframe the conversation on worker engagement, not just as an essential prerequisite to worker protection but also as an opportunity for companies to benefit from improved innovation processes and ultimately, improved productivity.

An Equitable AI Transition

For businesses and working people alike to feel the benefits of AI, we need new narratives that center worker voices, narratives that rebalance current power dynamics towards working people and away from a small number of tech companies. With AI, we have the opportunity to enhance livelihoods, reduce socioeconomic disparities, and simultaneously, improve business outcomes. To ensure this becomes a reality, we need to tell new stories. Stories that redefine who is an expert and who needs to be consulted in the design, development, and deployment of AI in the workplace. For all of us to benefit from the integration of AI at work, we need to strengthen the voice of workers, all
workers.

Read more stories by Alex Ash.