AI in the Edit Suite
INFLUENCER: Big Chop's Matthew Felstead says editors' jobs aren't under threat from the robots... for now
Last year, the Internet was awash with news proclaiming the beginning of the end for editors. Adobe had created a fast, efficient editing software that could produce edits just as professionally as trained editors, in half the time.
Aside from the obvious flaws in the editing software in terms of dialogue clipping, continuity and performance selection, the AI is obviously a fairly impressive tool. Used in the correct environment it could be quite beneficial. I just don’t think its place is in an edit house, for a number of reasons.
Firstly, the organising, labelling and compiling of rushes is a valuable learning tool for a junior assistant, as it gives them first-hand experience of assessing the performance of the actors and learning about the software they are using. For more experienced assistants, the chance to cut a rough edit for the editor and benefit from their feedback is a vital part of transitioning from assistant to editor.
Secondly, there is obviously the human element to editing. The physical act of sitting through the rushes and watching every take that the narrator of the clip calls ‘tedious’, is actually of paramount importance for an editor. It enables them to know exactly what they are working with and make a note of the variations in each take. This obviously helps them to make the best decisions on what will work in the edit and to be able to identify any performance the director, agency or client may request. This brings us to the biggest failing of the AI editing. It’s apparent inability to physically be in a room with director, agency or client and respond to their requests where shots and edits can be switched out or moved by frames to change the feel and structure.
This brings us to the third and biggest failing of the AI editing software: its inability to communicate directly with the client or creative team. This means it cannot respond to their requests for shots and edits to be switched out or moved by frames to change the feel and structure, nor can it share its opinion and expertise on whether edits are working or not. Negotiating and interpreting the changing and sometimes competing ideas of so many different creative minds and personalities is a humanised skill developed through years of doing the job.
I am confident that the number of idioms the AI would need inputted to successfully identify a specific take would be competitive with the time it would take human editor to craft this manually.
A common misconception about editing that I believe the AI in its current form makes, is that it’s a simple case of finding the best takes and performances and putting them in the right order: in other words, that editing is a technical job. In actual fact, it’s not as simple as that. Each shot has a direct relationship with the one before it and the one after it and it’s about finding the perfect balance of these shots that can lift an edit. It’s a skill and craft that takes years to master.
Furthermore, editing, like all aspects of filmmaking is a craft, involving creativity, time, years of training and, most importantly, collaboration. I think this is best achieved by the team involved with the project communicating face to face.
For the moment, anyway.