A documentary about training AI on real people raises questions it refuses to answer

Marc Isaacs' new film 'Synthetic Sincerity' blurs fact and fiction to explore how AI learns from human faces and stories. Critics say it skips the hard questions it promises to ask.

AI2Day NewsdeskAI-assistedPublished Updated Editor: Lee Brown3 min read
Illustration: a smartphone screen displaying a grid of generic social media advertisement tiles
Illustration made with AI. Not a photograph of the events described.
Share

Key points

  • Marc Isaacs, a British documentary maker, released a new film in 2025 called Synthetic Sincerity that mixes real documentary footage with scripted fiction.
  • The film depicts a fake AI research lab using real people from Isaacs' previous films to train artificial intelligence software.
  • Romanian actor Ilinca Manolache's face was digitally transformed into an AI avatar character without the film showing how or whether she consented.
  • The Guardian called the film shallow and self-aware in a way that undermines its own subject matter.
  • Who owns a person's face when AI trains on it is the film's central question, and it goes largely unanswered.

Marc Isaacs built his reputation making intimate documentaries about real people. So when he turned his camera on artificial intelligence, the expectation was that he'd do what he does best: get close, ask the uncomfortable question, sit with the answer.

Synthetic Sincerity doesn't quite do that.

The film imagines Isaacs licensing his own past subjects, real individuals from his earlier acclaimed work, to a fictional lab called Synthetic Sincerity at the equally fictional University of Southern England. Its stated goal: training AI software to generate convincing human figures for a screen. AI systems learn to imitate faces and personalities by studying vast libraries of real human material, often without those people knowing. That's a genuinely urgent concern right now, as Australian artists fighting for similar protections showed us just days ago.

The lab's researchers are played by actors. Lebanese independent film-maker Lynn El Safah appears among them. Isaacs has scripted, often funny conversations with an AI avatar, a glowing digital face recalling Max Headroom, the sardonic computer-generated TV personality from the 1980s. That avatar's face is modelled on Romanian actor Ilinca Manolache, known from director Radu Jude's film Do Not Expect Too Much from the End of the World.

Here's where it stumbles.

Manolache's real face became the AI character's digital mask, and the film never shows how she was approached or what she agreed to. For a documentary positioning itself as an interrogation of AI and identity, that silence is telling. The very process under examination happens off camera.

As The Guardian noted, the result is a project that's "itself exasperatingly artificial." A film about AI sincerity that withholds its own behind-the-scenes truth.

Does any of this matter to ordinary people?

Directly, yes. Across entertainment and beyond, real people's faces and personal histories are being used to train AI systems, sometimes without clear consent and rarely with payment. Synthetic Sincerity gestures at that problem without pushing hard enough to show what it looks like in practice.

For anyone who's watched the consent debate play out in courtrooms and parliaments this year, the film's reticence isn't just an artistic choice. It's a missed opportunity. Worth watching if you're already interested in how AI and identity collide. Less useful if you want answers.

© 2026 AI2Day