AI vs Humans in Video
Algorithms are not sentient beings and don’t have emotions. They can do a semi-decent job at identifying emotional cues in videos by relying on speech and vision. Algorithms are (still) horrible at detecting wisdom and subtle humor. They are also in their infancy when it comes to adding magical special effects. Humans can do a far better job at judging portions of a video for emotional content and for adding special effects.
The domain Parmonic operates in is informational videos. These are videos heavy in their information content. Algorithms are very good at “watching” millions of hours of boring videos and classify content, match it with domain dictionaries, recommend topics. They can learn about any industry/vertical/company or knowledge area very quickly (e.g. we can literally train our engine in seconds about certain topics that would take years for a human to learn).
Under the core of video editing/transformation lies computer code. Actions that one might take in a video editing software like Adobe Premier get converted to lines of code that the computer executes to add those clouds and bubbles. Every single task gets converted to lines of code. Algorithms are very good at automating programmatic tasks in video transformation. While a human can click the same button 20 times, an algorithm can “click” it infinitely and as-needed.
S. Somasegar (previously CVP at Microsoft where I also spent a decade) talks about 3 different types of intelligent applications – Automators, Augmentors, and Avant Garde.
For 90% of video content, the automation offered by AI in grunt and programmatic work makes it a huge win. It is faster, easier, cheaper.
For the other 10%, the augmentation it provides so humans can spend more of their time on those emotion-tingling special effects or picking subtle humor, makes it extremely compelling.
Magically “snackifying” videos with AI and minimal human supervision – that’s Avant Garde.