AI filmmaking tools look revolutionary until you see the cost, chaos, and lack of control behind those perfect 8-second clips.

Filmmaking has always been about intent. Every frame, every shadow, and every movement is a deliberate choice. But lately, we’re being told that this entire craft can be boiled down to a text prompt and a thumbnail. While AI developers are becoming billionaires overnight, those of us in the trenches are finding a much messier reality.

The “Credit Burn” Reality

We’ve all seen the viral clips on Twitter, but ask any creator what it actually took to get those eight seconds of “magic.” They won’t tell you about the hundreds of dollars in credits they burned or the hours spent shouting at a prompt box that refused to listen.

Right now, using AI video tools feels less like filmmaking and more like gambling. You put your money in, pull the lever, and pray the AI doesn’t hallucinate a third arm or turn your protagonist into a puddle of pixels. In their rush to compete with each other, companies like Kling, Sora, and Veo are chasing “version numbers” instead of fixing the broken foundations.

Why Control Matters

The “Pro” versions aren’t much better. Even with a perfect prompt, these tools still “do their own thing.” For a professional, that’s a dealbreaker.

• Traditional VFX and CGI? They’re hard work, but they’re reliable.

• DaVinci Resolve? It does exactly what I tell it to do.

• AI Tools? It’s a coin toss every time you hit “generate.”

I’m not an AI hater. I see the potential for AI to handle the boring stuff the rotoscoping, the denoising, the grunt work. But right now, the marketing is lying to us. They’re selling “cinematic revolutions” while delivering expensive experiments.

The disconnect isn’t just technical; it’s emotional. As filmmakers, we pour our souls into the “geometry” of a scene. We obsess over the way a light hit a lens, the shots or the specific pacing of a character’s gait. When an AI developer tells us a text prompt is “the new cinematography,” it feels like an insult to the craft of filmmaking. It ignores the thousands of micro decisions that make a shot feel human.

Right now, using these tools feels like trying to direct a cinematographer who refuses to look at your storyboard and speaks a different language. You ask for a close-up of a mourning mother, and the AI gives you a neon-lit cyberpunk city because “cyberpunk” is trending in its training data. That’s not “innovation” that’s a lack of basic utility.

We’re tired of the “Coming Soon” trailers for tools like Sora, Veo 3,  OpenAI’s Sora, Google’s Veo, Kling AI,  Runway Gen-4 , Luma Dream Machine, Pika, Hailuo AI, and Adobe that look flawless in a controlled demo but fall apart the second a paying user tries to recreate them. Stop chasing the next viral Twitter aka. X clip and start fixing the temporal flickering that ruins every third render. We don’t want magic; we want a tool that actually works when the deadline is staring us in the face.

A Message to the Developers

If you want us to take your tools seriously, start acting like you care about the people using them:

• Stop ignoring the bad reviews. We aren’t complaining for fun; we’re complaining because we want the tech to work.

• Fix the hallucinations before adding “features.” I don’t need a new style filter if the character’s face still melts halfway through the shot.

• Build a support team. If I’m spending hundreds of dollars/ British pounds/ Euros a month, I shouldn’t be met with silence when the tool fails.

The Bottom Line: Transparency builds trust. Be honest about what your AI can’t do yet. We’d rather have a tool that does three things perfectly than a tool that promises the world and delivers a hallucination.

Explore Maria Johnsen’s fiction books