How real is real? AI and documentary reality
Dr Shweta Kishore is a lecturer and the lead of screen and media at RMIT University's school of media and communication. She explores the balance between innovation and authenticity in documentary filmmaking.
Artificial Intelligence (AI) is transforming every aspect of cinema from production to distribution and audience engagement. But it is not without its challenges, particularly in the case of documentary film which relies on values of truth, authenticity and facts.
While artistic creativity and selection are key to documentary representation, recent releases such as About a Hero (2024), Another Body (2023), Welcome to Chechnya (2020), Roadrunner (2021) highlight the ethical dilemmas AI introduces, including its environmental impact.
In adjacent fields of narrative cinema and video games, AI is already transforming reality into imaginative screen worlds. Games like Block Party create human avatar Non-Player Characters based on real documentary subjects. In documentary, AI is entangled with processes, from research – summarising online content using ChatGPT, digging into archives, transcription and translation of interviews, to creative functions such as filling gaps in footage and the editing of short visual sequences that bypass human editors.
The use of AI tools to, for instance, streamline, enhance, improve, clean, and organise documentary materials and processes continues, but it is the generative aspects of AI that cause ethical concerns. For one, researchers have noted the presence of real-world stereotypes and social biases in Large Language Models during text generation which also puts image generation into question.
Dr Shweta Kishore
AI and Questions of Authenticity
Ethical inquiries in documentary are not new and filmmakers are often scrutinised for choices that create doubt about the authenticity of documentary content, subjects and claims. However, accelerating use of AI technologies has intensified concerns around documentary’s contract to represent an authentic and verifiable reality.
For example, Roadrunner‘s use of deepfake technology to simulate Anthony Bourdain’s voice posthumously sparked heated debates about the place of synthetic media in non-fiction film. The fundamental question here underlines the ethical responsibilities of filmmakers towards audiences – an unspoken understanding that documentary sound and image is a representation of an existential reality, with transparency the key to audience trust.
In 2023, members of Archival Producer Alliance, who deeply understand the value of authenticity, trust and transparency, created a “Best Practices Guidelines” focusing on Generative AI in documentary. The guidelines distinguish between AI’s standard uses for tasks such as logging, transcription, or minor audio-visual corrections and generative AI’s more contentious capabilities to create new material and/or alter existing material which can mislead audiences. They call for strict regulation around the representation of ‘simulated’ humans urging filmmakers to take extra caution by acquiring consent of all those represented, managing cultural sensitivities and maintaining the sanctity of the historical record.
Anthony Bourdain
AI and Artistic Vision
Generative AI presents huge opportunities to filmmakers to access sensitive situations where participant care is fundamental. In Welcome to Chechnya, David France uses generative AI to create deepfake ‘masks’ to protect the identity of vulnerable LGBTQ participants, fleeing homophobic persecution in Chechnya. Similarly, in Another Body, directors Reuben Hamlyn and Sophie Compton use deepfake to maintain the privacy of an already traumatised young female participant. Unlike traditional solutions like silhouette lighting, the use of AI, retains emotional ‘presence’ attached to the human face while anonymising the participant. In Welcome to Chechnya deepfake is a narrative device that spotlights existential dangers, “melting” off when activist Maxim Lapunov decides to publicly speak out. AI use here elegantly balances artistic expression with ethical responsibilities towards both participants and audiences as well as documentary’s social responsibilities.
AI’s exploratory potential spans technical innovation to rethink the artistic propositions of filmmaking. Casting a reflective look inwards is crucial to the development of an art form. About A Hero (Piotr Winiewicz, 2024) audaciously tackles these questions with Kaspar, a custom-built AI ‘trained’ on Werner Herzog’s voice overs, writings and interviews to craft a signature Herzog screenplay. Juxtaposing the AI generated script with real interviewees, including Stephen Fry, opens a fascinating gap to probe Herzog’s style and indeed broader notions of ‘style and ‘authorship’ – a foundational aspect of our love for cinema.
However, the ethical implication of AI extends beyond the screen into the physical worlds that we inhabit. Researchers have identified urgent environmental issues associated with the energy consumption of LLM’s, including water for cooling systems and training needs. A recent MIT study finds that Image generation is the most energy and carbon-intensive AI-based task with a single AI generated image requiring as much energy as charging a smartphone. As generative AI tools like Sora and Runway gain popularity, its crucial to consider the environmental costs of these operations. Filmmakers must ask themselves – are they really aware of the hidden environmental costs of AI systems? Moving forward, transparency, around AI’s environmental footprint will be key to making choices that are both artistic and ethical.
Keep up to date with the latest in media and marketing
