CNET's Aaron Pruner walked into the world's biggest AI in media conference hoping to update his skepticism. The labor and environmental questions were the ones no one on the backlot wanted to host.
CNET's Aaron Pruner, a SAG-AFTRA member and recent striker, walked into AI on the Lot, billed as the world's biggest conference focused on AI in media, hoping to update his skepticism. Roughly 2,500 people had gathered on the Amazon MGM Studios backlot in Culver City for the event. He came out, by his own account, more conflicted than when he arrived. That is a more useful starting position than either the vendor pitch decks or the strike-line manifestos, and it is worth treating the conference the same way Pruner did: as a working map of where the AI-in-film conversation actually is, not a verdict on where it is going.
The artifacts are no longer hypothetical. Hell Grind, an AI-assisted feature, screened at Cannes. Dream of Violets became the first AI-made feature to play Tribeca. These are no longer slideware or vendor demos. They are submissions that real festival programmers decided to program, which is a different register than anything coming out of the keynote hall. The conversation has moved past prediction into curation, and that shift is worth holding in mind while sorting the rest of the noise.
The first disconnect Pruner flags is product hype versus production reality. Vendors on the show floor were pitching tooling that would compress previsualization, automate rotoscoping, and generate B-roll at speed. The promise is real, and some of it is shipping into real pipelines. But the version of the future being sold in the keynote hall ran several steps ahead of what assistant directors, VFX leads, and post-production supervisors described in hallway conversations. Tools that vendors say will replace a workflow still require the workflow's experts to operate them at production quality. That gap is not a bug to be fixed in a future release. It is the actual shape of the transition.
The second disconnect is techno-optimism versus worker fear. The conference programming was heavy on creators who had built careers around AI tools, and lighter on the people whose jobs those tools are most likely to compress. That is not a conspiracy. It is how vendor-financed conferences tend to program. The effect, though, is to make the optimistic case feel like consensus and the worried case feel like Luddism. A more honest room would have included organized labor, laid-off production coordinators, and the support staff who keep a set running. The conference was held on a lot where some of those workers are still in dispute over the very automation being celebrated indoors.
The third disconnect, and the one Pruner presses hardest on, is the absence. Labor displacement and environmental cost were the two critiques that almost no panel wanted to host. Powering generative video at scale is materially expensive, and the consolidation of creative decisions into fewer hands is a structural question, not a vibes question. Neither topic got the airtime that the latest text-to-video model did. Whether that silence reflects a real belief that the questions are settled, or a strategic choice to keep the room friendly to investors, is itself a question worth asking.
A working map for the reader who has to make decisions now, rather than predictions, looks like this. First: separate assistive tools, VFX pipelines, generative video, and full generative features. They are different products with different timelines and different labor impacts, and the trade press often blurs them for narrative convenience. Second: ask any vendor pitch what the workflow it replaces actually looks like today, and how the displaced worker's expertise gets folded into the new pipeline, if at all. Third: when a conference is held near a working backlot, ask who is not in the room, and what is not on the stage.
The honest position is not that AI is saving Hollywood or killing it. The honest position is that the pipeline is being reorganized in real time, that the people most affected are not the people most quoted, and that the questions worth asking are the ones the industry has so far declined to host. Pruner went in a skeptic and came out conflicted. For working creatives and curious readers trying to navigate the shift, that is a better place to start than a press release.