AI in Audiovisual: Navigating Ethics & Responsibility
Here at Bendita Filmes, we experience this firsthand every day. Since 2010, we’ve witnessed the evolution of Brazilian audiovisual production, and now, AI is the topic dominating backstage conversations. It promises to streamline processes, optimize costs, and unlock new creative avenues. But, like any revolution, we must pause and consider: at what cost? Integrating AI without critical evaluation can lead to more problems than solutions. That’s what we want to discuss today, drawing from our experience on the production frontlines, facing practical challenges and contemplating long-term implications.
Foto de Google DeepMind no Pexels
Algorithmic Bias: Society's Broken Mirror
One of the most delicate points in using AI in audiovisual is algorithmic bias. AI tools learn from data. If this data reflects historical prejudices – racism, sexism, etc. – the AI will inevitably perpetuate and even amplify these distortions in everything it generates. We see this happening across various fronts, and correction isn't straightforward. Blaming the algorithm alone isn't enough; responsibility lies with the data that feeds it and the human curation that should, but doesn't always, provide critical analysis.
What We Observe in Practice:
In generating characters for animations, for instance, we've noticed that some image generators, when asked to create a "tech professional," tend to present a more homogenous profile in terms of ethnicity and gender, if the training data wasn't carefully balanced. This isn't a flaw of the AI itself but a direct reflection of pre-existing data. Today, we have access to more sophisticated tools for adjusting diversity in prompts, but even so, it's the creator's attentive eye that ensures we aren't perpetuating stereotypes. For example, in a project for an educational sector client, we had to refine image generation prompts to ensure that the "scientists" and "teachers" represented Brazilian diversity – something a generic generator, without this specific instruction, wouldn't naturally do.
Trade-offs and What Tutorials Don't Tell You:
The catch here is that to combat these biases, simply using the tool "out of the box" isn't enough. It requires an investment of time to refine prompts, carefully select training datasets (if you have control over them), or, more commonly, conduct rigorous post-generation curation. This demands time and expertise, factors not always accounted for in tight budgets for AI video projects. Beginners often believe AI "does it all" and end up delivering a product that, unintentionally, reinforces what we most want to combat.
Authorship and Copyright: Who Owns the Idea?
This is the million-dollar question in the audiovisual market today. When an AI generates a script, a storyboard, or even a soundtrack, who holds the copyright? The tool's developer? The user who wrote the prompt? The AI itself? Legislation is still catching up, and the legal landscape is a minefield.
What We Are Experiencing:
In a recent experimental short film project, we used AI to generate various script twist options. We selected the best ideas, but the work of weaving a cohesive narrative with emotion and a consistent character arc was 100% human. The question is: how do we attribute credit? In our case, we opted to mention "with AI assistance for script ideation" in the technical specifications, but for the final audience, the credit belongs to the human scriptwriters. The difficulty arises when the AI generates a complete piece, and distribution platforms are already asking for clarity on content origin.
Nuances and the Reality:
In many jurisdictions, copyright law is clear about the necessity of human authorship. This means, in practice, AI is an assistive tool. The human creator is the one who signs the work, being responsible for its integrity and originality. The major challenge for independent producers and small studios like ours is navigating this legal limbo without taking unnecessary risks. Larger companies are investing in specific AI legal consultations, but for most, caution and clarity in documentation are essential.
Transparency: The Bridge of Trust with the Audience
A lack of transparency about AI use can erode audience trust. Today, the line between the real and the synthetic is increasingly blurred. Disclosing when and how AI was used is fundamental, especially in content that aims for verisimilitude or is presented as purely original.
Our Commitment and Practical Examples:
We at Bendita Filmes have always championed honesty in communication with our clients and the public. If an animation was created with AI assistance for generating background movements, or if a narration was synthesized, we believe it's important to communicate that. For instance, in an institutional video where we needed a presenter avatar whose speech could be quickly altered, we used an AI tool for the voice. We made this clear to the client, and the result was positive, as they could update the script without re-recording everything. Transparency avoided unrealistic expectations and built a trusting relationship.
The "Catch" and Trends:
The big question is: to what extent should we disclose AI usage? If we use an AI-powered video editor to suggest cuts, or a color grading software that learns our styles, does this need to be mentioned? Current trends point towards a need for clearer identification, especially in content that could be mistaken for factual reporting or 100% human creation. Digital watermarking tools and metadata for AI-generated content are gaining traction, and platforms will likely start requiring these markers. For us, this is a sign of market maturity, where quality and creative intent will be increasingly valued, regardless of the tool used.
Conclusion: Human Creativity as the Guiding Star
Artificial Intelligence is undoubtedly a powerful tool for audiovisual production today. It can expand horizons, optimize workflows, and even democratize access to resources that were once the privilege of large studios. However, as a production company with years of experience, our view is that AI should serve human creativity, not replace it. The ethical issues – bias, authorship, and transparency – are barriers that we can only overcome with critical awareness, responsibility, and an unwavering commitment to authenticity.
At Bendita Filmes, we see AI as an ally that enhances our capabilities. Our focus remains on telling impactful stories with emotion and truth, using the most advanced tools available to deliver high-quality educational content. The future of audiovisual production is collaborative: human and machine, together, but with human intelligence and sensibility always guiding the way.
About Bendita Filmes
Bendita Filmes is a full-service video production company and agency based in São Paulo, Brazil, specializing in:
Corporate Videos · YouTube Production · Instagram & TikTok · Advertising Campaigns · Event Filming · VFX & Animation · Live Streaming · AI Video Production · Photography
