Taming Technology episode 1 - Gunnar Ásgeir Ásgeirsson

From projectionist at Bíó Paradís to tech entrepreneur

Gunnar began his career as a projectionist and experienced firsthand the revolution when cinemas moved from traditional film to digital screenings (DCP – Digital Cinema Package). This change required new knowledge of computers and networks and sparked his interest in the technology behind the scenes.

This interest led him to found the company Sinilab, which specializes in creating screening packages for films and television shows. A large part of that work is subtitling, where accuracy and contextual understanding are paramount.

The challenges of artificial intelligence in Icelandic subtitling

Although artificial intelligence can translate text between languages at lightning speed, the process is more complex than many think, especially when it comes to Icelandic. In our conversation, Gunnar described his experiments with using AI to translate films and pointed out several key challenges:

• Context is everything: The AI must understand what is happening in the film. Words can have different meanings depending on the emotional context or the characters' body language.
• Gendered language: Icelandic has gendered inflections. The AI needs to be able to identify who is speaking (male, female, or other) for the translation to be grammatically correct.
• Security and confidentiality: Gunnar places a strong emphasis on working with sensitive material, such as unreleased films, on computers that are not connected to the internet. Therefore, he primarily uses open-source solutions that he can run on his own hardware.

Security and Open Source: Why Gunnar works offline

Gunnar explained how solutions that run without an internet connection ensure that sensitive data does not leak. This is a fundamental principle when collaborating with producers and demonstrates a responsible approach to using technology where privacy and intellectual property rights are at stake.

Main points from the episode

• The technological revolution in cinemas shifted responsibility from projectionists to technicians.
• The biggest challenge for artificial intelligence in subtitling is understanding context, emotions, and cultural nuances, not just translating words.
• Icelandic language technology faces unique challenges due to its gendered language.
• The use of open-source and offline tools is key to ensuring data security in the film industry.
• Continuous learning and curiosity are necessary to keep up with the rapid pace of technological development, similar to previous revolutions like the printing press and the internet.

Quotes from Gunnar

"The translation itself was perfectly fine... but it just didn't fit with what was happening in the movie... the context needs to be translated too."

Conclusion

My interview with Gunnar Ásgeirsson provided a unique insight into the practical applications and challenges of artificial intelligence in creative industries. It shows that although the technology is powerful, it can never fully replace human expertise, context, and responsibility. His story is a reminder of the importance of being curious, adaptable, and aware of the ethical aspects of technology.