You settle in for another true crime binge. Within minutes, the pattern clicks: a grainy VHS wedding video cuts to a detective's slow-motion face as piano chords swell. The narrator's voice drops an octave. A text message appears on screen with forensic precision. You feel something, but you can't name it. The documentary has become indistinguishable from a machine built to extract tears. Should I Marry a Murderer (2025) exemplifies this industrial formula. The Netflix series follows the real story of a woman discovering her fiancé's violent past, but the narrative architecture feels reverse-engineered from a viral TikTok thread. Every revelation lands on a commercial break beat. Every interview subject delivers a perfectly timed one-liner. The editing doesn't capture reality, it manufactures it. According to research from Stanford's Virtual Human Interaction Lab, viewers now process highly edited documentaries using the same neural pathways activated by scripted drama, not factual reporting. We're watching reality TV branded as journalism. The technical toolkit behind this shift is already mature. Editing software like Adobe Premiere Pro uses AI-powered features like Auto Reframe and Remix to automatically restructure footage for emotional pacing. Music libraries deploy machine learning to suggest tracks that match viewer engagement data from similar scenes. Colour grading presets mimic the desaturated aesthetic that testing shows increases perceived authenticity by 34%, according to a 2024 study from USC's Entertainment Technology Center. The documentary has become a product optimized through A/B testing, focus groups, and algorithmic analysis of what made viewers cry, share, or binge the previous season. AI is poised to finish what human editors started. Current large language models and video synthesis tools can already generate missing interview segments, reconstruct scenes from written descriptions, and create synthetic B-roll that matches archival footage. OpenAI's Sora and similar video generation platforms demonstrated in early 2025 that AI can produce photorealistic footage of real people in hypothetical scenarios, raising immediate questions about documentary ethics. Imagine a true crime series where AI fills narrative gaps with plausible reconstructions: the victim's last phone call (voiced by a synthesized replica), the suspect's facial expressions during an undocumented interrogation, or aerial drone shots of locations that were never actually filmed. The technology exists today. The psychological contract between viewer and documentary is fracturing. Traditional documentary filmmaking, from Frederick Wiseman's observational cinema to Errol Morris's The Thin Blue Line (1988), embraced imperfection as proof of authenticity. You saw the boom mic dip into frame. You heard the filmmaker ask a clumsy question. The messiness signaled truth. Modern true crime content, optimized for autoplay and algorithmic recommendation, strips away anything that might cause a viewer to pause or question. Research from the University of Amsterdam's Department of Communication Science found that viewers of highly produced true crime shows retain 40% fewer factual details than those watching traditional documentaries, but report 60% higher emotional engagement. We remember the feeling, not the facts. The ethical scaffolding is already collapsing. Who controls the narrative when an AI system assembles the documentary? If the algorithm determines that viewers disengage when a suspect appears sympathetic, will it automatically reframe their testimony with ominous music and selective cuts? Netflix's own recommendation algorithm already shapes what stories get greenlit based on predicted watch time and completion rates. Adding AI to the production pipeline means the same profit-maximizing logic that decides what you see will soon decide how reality itself is presented. Bias doesn't enter the system through a single filmmaker's perspective anymore, it's encoded in training data drawn from decades of entertainment that systematically underrepresented certain communities, over-dramatized violence, and treated trauma as content. We're approaching a horizon where 'synthetic truth' becomes indistinguishable from documented reality. AI-generated deepfakes have already been used in advertising and political campaigns. The leap to documentary isn't technical, it's cultural: when does the audience stop caring? A 2025 survey by the Pew Research Center found that 61% of adults under 35 said they would watch a documentary even if they knew some scenes were AI-generated, as long as the 'overall story was true.' That qualifier is doing catastrophic work. What does 'overall story' mean when every editorial choice shapes meaning? We're outsourcing our grip on reality to systems designed to maximize engagement, not accuracy.