The Clock, Christian Marclay’s video installation, mixes thousands of cinematic scenes that present time through clocks (and watches) and events in movies around the concept of time, creating a 24-hour montage from them. Marclay’s video creation functions as a huge “clock” synchronized with local time at the geographical point where it is on display. The installation creates tension between cinematic and real-time, enhances the viewing experience and challenges the boundaries of the exhibit.

Christian Marclay collected different behaviors and moments around the dimension of time in cinema. Marclay’s installation crosses all types of cinema and looks at many periods in cinema. The dimension of time in Marclay’s work also presents developments in the technology and design of clocks on the cinematic sequence of time corresponding to the present time in the period that the cinematic work was created. Various behavioral structures can be observed around the concept of time in Marclay’s work. The body language, dialogues and monologues of the actors convey different feelings around the concept of time. Some of these behaviors relate to daily dreariness, while others are disturbing and stressful. The Clock shows hundreds of cuts of body language related to time, desire, pressure, fear, mystery, and so on. The cinematic interactions portray different behaviors toward the same theme. This is not just about verbal interaction around time, but also about body language around the concept of time.

How does this relate to a model for artificial intelligence?

Marclay created a cinematic database around the world of time. That database contains thousands of human (cinematic) behaviors around the concept of time, which can be entered into a model that has a sole purpose of identifying and creating human behavior around the concept of time. This model can be reproduced for other cinematic experiences and behaviors around other events. and these cinematic behaviors can add a human layer to human-computer interaction. The model can also be used to identify and understand the human space through those cinematic interactions, or to emulate them in order to add a human layer to the software.

One of the major challenges in creating a model for artificial intelligence is the need for a large amount of data, segmenting, indexing the data according to different fields or categories, and “training” the data to create a model that can be used for future processes or interactions. The cinematic world meets all these needs, and the Marclay’s Clock Model, as I call it, can serve as an open-source platform for the purpose of creating human interactions by the machine, or for identifying human behaviors while learning from the cinematic world. The model can be used for various applications that require computer-human interaction, such as ordering a hotel room meal or receiving help and support in various matters. To do this, hundreds of cinematic interactions concerning the same subject can be fed into the ‘Clock Model’, which could then derive artificial insights regarding that situation.

Cinematic data is already well-indexed and contains a very broad database for many human interactions, which are seemingly interwoven on the fictional and realistic level. It also includes different cultures and languages ​that could be used to derive models that differentiate between specific local norms and cultures. By so doing, it could include culture-dependent gestures and avoid reactions or behaviors assigned to one culture or another. Naturally, the cinematic world can create a mostly biased behavioral model that is basically unrealistic. For example, if the data fed into the model is mainly of violent cinematic scenes, then the model will be able to emulate and identify mainly violent behaviors.

There are many commercial companies that have already catalogued the cinematic world, so this is very similar to as if we were to catalog hundreds of thousands of different behaviors from the real world. These companies use artificial intelligence to automatically understand cinematic context, events and occurrences. This information can be used for the Clock Model. Of course, we need to distinguish here between genres, in order not to create a distorted model, but there are enough cinematic genres to shed light on the non-cinematic world. Models that imitate different worlds can always be made to create a different experience from the normative experience, by examining the same interactions in different cinema genres such as psychological thrillers, action and science fiction movies. By so doing, we can enable the system to have a variety of reactions, which can sometimes be humorous, mysterious or disturbing.

This proposal is a general introduction to the Clock Model. I would be delighted if people who work in artificial intelligence showed interest in taking the idea and developing a learning-based interface from the cinematic world together. It may be possible to build entertaining human bot interactions from this, based on various cinematic genres that would form a humorous connection between the machine and the user, or to try to take this model and link it to non-entertainment worlds, such as receiving services, resolving financial issues and so on.