AI Narratives
- Written byRob Eagle
- Published date 21 July 2026
Share story
Turning anxiety into agency: Narratives for a different AI future
Artificial intelligence is often presented through binary extremes. Depending on who is telling the story, AI is either humanity's salvation or its downfall. It will cure every disease, transform productivity and unlock creativity—or destroy jobs, spread misinformation and threaten democracy.
These types of stories leave many of us feeling powerless. But what if having such extreme narratives on AI is part of the problem?
At Sheffield DocFest, the panel on Turning Anxiety into Agency: Narratives for a Different AI Future examined how the stories we tell about AI actively shape who holds power over it and whether the audience feels they have any influence over its future applications.
Moderated by Francesca Panetta, Director of the AKO Storytelling Institute at the University of the Arts London, the discussion brought together documentary producer Diane Becker (The AI Doc: Or How I Became an Apocaloptimist), immersive director and producer Shehani Fernando (Vocalize), AI entrepreneur Ángel Maldonado (Empathy.AI) and Elise Tillet-Dagousset, Campaigns & Media Lead for Europe at Luminate.
The panel challenged storytellers to shift away from extreme projections of what AI might be and instead to tell stories that help the public understand that the future isn’t binary and dystopian AI adoption isn't inevitable. Stories that remind us of our agency to help to shape our future with AI.
Narrative change and how stories of AI accumulate
Cultural narratives are shaped by a dynamic constellation of individual stories. Narratives are the dominant frameworks we see repeated in ideas, metaphors and assumptions that shape how society understands a topic, such as AI. As Panetta explained, narratives can be mapped through media, films and public discourse. They shape our perceptions of risk, responsibility and possibility.
Elise Tillet-Dagousset highlighted several dominant narratives currently shaping public understandings of AI:
- The ‘doom vs salvation’ binary - AI will either destroy humanity or save it.
- The ‘superintelligence inevitability’ story - the future is already decided.
- Anthropomorphised AI - systems described as if they ‘think’, ‘feel’ or act independently.
- Tech neutrality - the belief that technology is objective, not shaped by human decisions or incentives.
In the UK, according to a recent survey of nearly 3,000 UK households, 69% of the population is concerned with the direction that AI is heading (with only 19% self-declaring as ‘excited’). 80% don’t trust the government to manage it, and even more want stronger government action. Generally, people already see AI deployed in their daily lives and don’t feel that it is designed for their benefit; rather, they view the inevitable growth of AI as driven by profit for shareholders of the main tech companies.
Research published by USC’s Norman Lear Center in 2025 demonstrated that social media (and the algorithms developed to influence users’ behaviours) is often framed as an unstoppable force or sentient actor. However, what results is that responsibility disappears, harms to users become abstract and our feeling of agency in the matter dissolves. For storytellers, this illustrates how their work shapes the narrative landscape, not just reflects it.
Helpfully, a team of researchers between Europe and the US have recently offered guidance on de-anthropomorphising language to describe AI technologies. They outline how the language journalists, companies and scholars use to characterise technology influences how it is perceived, deployed and trusted. They then present concrete strategies for informing and empowering audiences by using appropriate terminology and avoiding inflated claims.
Moving beyond fear
A recurring theme across the panel was the rejection of binary positions. Diane Becker reflected on how the documentary she has produced, The AI Doc: Or How I Became an Apocaloptimist, attempts to avoid simplistic answers. The film follows co-director Daniel Roher, who is preparing for the birth of his first child and trying to understand what kind of world AI is creating. The filmmakers recognised early on that chasing every new AI development would be impossible because the technology was moving too quickly. Instead, they focused on the human questions beneath the technology.
Their approach to storytelling meant they avoided extremes, created space for nuance and left space for the audience to feel a sense of agency, rather than hopelessness. The result, Becker explained, was a more balanced discussion about AI.
Similarly, the immersive experience Vocalize avoids lecturing participants about the dangers of AI. It works as a media literacy tool by inviting the audience into an 'entertaining' experience with a trusted narrator in the form of Stephen Fry. There, they unknowingly train an AI model on their own voice before hearing themselves speak words they never actually said. The creators hope that, by experiencing firsthand how their voice could be cloned and used to spread misinformation, the audience can build resilience against synthetic media in a more effective and lasting way. In social psychologist Melisa Basol's evaluation of the project, ‘Participants shifted from passive awareness to embodied insight, with 100% of those who responded reporting an increase in awareness of cloning that made risks feel immediate and personal.’
The third project featured on the panel was the documentary Molly Vs THE MACHINES, the story of a father’s quest to understand how social media contributed to his daughter's death and his battle against the Silicon Valley corporations he and other campaigners hold responsible. Broadcast in March 2026 on Channel 4, Tillet-Dagousset describes it as ‘not a tech story. It’s a story about power, social justice and humanity.’
These three projects demonstrate how using stories of emotion and lived experience can be more effective than an abstract explanation of the technology. As Tillet-Dagousset highlighted, too often technology-focused stories remove AI from its physical, visible and tangible contexts.
That is why Ángel Maldonado and his colleagues built the interactive chatbot for the Molly Vs THE MACHINES website, built with the transcript and background information for the film, as well as supporting research. They are transparent about the data they used to train the model for the chatbot; this also exposes how other AI tools obscure their sources when they are reliant on practices of non-consensual model training, such as data scraping.
Six key takeaways for changing the narrative on AI
The discussion offered a range of practical guidance for storytellers interested in changing narratives on AI. These offer principles that can be applied to many issues involving power, technology and social change.
1. Reassign agency by naming human actors.
Avoid language that humanises AI (e.g. ‘AI decided’ or ‘AI thinks’). Storytellers should instead focus on the real companies, developers and decision-makers behind the technology. This helps to show the human labour developing the systems, as with Empathy.AI’s work. This simple linguistic shift emphasises accountability for those in power and makes change imaginable for the audience.
2. Deconstruct the ‘magic’ of AI.
AI is often framed as mysterious or unknowable. Maldonado’s work demonstrates the opposite by making AI physical, visible and tangible. For storytellers, this might mean visualising infrastructure, showing the processes that go into building large language models and revealing hidden labour. Demystification gives the audience a practical and critical understanding of how AI is built and works.
3. Put people (and the harms they experience) back in the frame.
Often, representations of AI focus on distant futures, which distract from real impacts happening now. Instead, storytellers ought to highlight harms like the workers labelling data, the communities living near data centres and the victims of deepfakes. Storytellers can do this through connecting individual stories to systemic drivers of harm, as the makers of The AI Doc and Molly Vs THE MACHINES approached their topics. Don’t just show harm—show what causes it.
4. Tell stories of resistance and possibility.
The media landscape is saturated with stories of both dystopia and hype. What’s missing are stories of collective resistance, examples of alternative systems and narratives where change is already happening. These don’t have to be utopian, but they should expand what audiences believe is possible. We see examples of this in Molly Vs THE MACHINES and the work of Empathy.AI.
5. Balance impact with pathways for action.
A core tension raised in the panel was how to avoid leaving audiences overwhelmed. Fernando described the need for a ‘decompression zone’ for Vocalize— a space for reflection, discussion and continued engagement after an emotional experience. Becker added that even simple prompts can spark awareness and agency. Ultimately, not every story needs a call-to-action for the audience, but it should leave room for action to exist.
6. Don’t overburden the individual.
One of the most important cautions from Tillet-Dagousset was that focusing too much on the individual responsibility of the audience can itself be misleading. Just as with climate change, systemic problems require systemic solutions. Storytellers can, instead, avoid implying audiences must ‘fix’ everything themselves, highlight the role of institutions, governments and corporations and show how public pressure can enable change.
From anxiety to agency
The panel returned repeatedly to the question of what agency actually looks like. This involves more than giving audiences a simple call to action and changing what they believe is possible.
Recently, a collective of AI awareness advocates published the AI Resist List, a major new effort to create a database of AI-critical movements from around the world. It’s a useful resource of organisations and campaigners with different approaches working to reshape the trajectory of AI development.
Reflecting on the session’s title, we can see a common challenge for storytellers wanting to shift narratives on AI. Anxiety is easy to spark, whereas inspiring agency is far more challenging. While fear can grab attention, meaningful change often comes from helping audiences to understand their place within a larger system and showing them where influence and accountability lie.
As storytellers have a unique role to play in shaping how society understands AI, they have a responsibility to do more than perpetuate existing extreme narratives. Stories can influence whether audiences feel powerless or empowered. The shift from anxiety to agency requires more than better technology. We need better stories about technology.
Written by Rob Eagle
Turning Anxiety into Agency: Narratives for a Different AI Future an AKO Storytelling Institute event for Sheffield DocFest
Speakers:
- Shehani Fernando (Immersive Producer/Director, Vocalize)
- Ángel Maldonado (Founder, Empathy.AI)
- Diane Becker (Producer, The AI Doc: Or How I Became an Apocaloptimist)
- Elise Tillet-Dagousset (Campaigns & Media Lead - Europe, Luminate)
Moderator:Francesca Panetta (Artist and Director of AKO Storytelling Institute, University of the Arts London)