Work is traditionally an objective act, structured by procedures, for the sake of reproducible results. Play on the other hand is experimental, unstructured, free roaming, with unexpected results. Both work and play form their own space, and the digital space of a computer make work and play possible at the same time. The multimedia capability of computers not only makes both making and consuming media possible, but also makes the domain of separate activities a singular process, more often called multi-tasking.

As computer operations have become more powerful, faster, and networked, the differences in labor and leisure on a computer have been blurred. For users, a computer is a multi-media machine that can shift into alternate modes of activity instantly. Things like work processors, media editing, internet searches, and file management or selection are interactions made possible by human command and computer operations. However, casual searches for content can be done at the same time as real work is happening. As much as we can do work on a computer, we can also use them to access information, articles, movies, music, and games.

We might listen to a podcast or have video streaming playing in the background or to the side of work driven programming. With so many actions and digital events at play on a computer, productivity on a computer can become murky. Yet, this framework of human to computer is already a basis for digital tools driven by human users.

Today’s digital toolkit is fixated with AI driven operations. Getting computers and information networks to generate unique text and image from data. If labor and play are interchangeable on computers, then AI better illustrates this mode of production with work that is done by playing with an AI model. When a specific user prompt generates unexpected results on screen, the result is indeed experimental, and the experience is novel for the viewer as they wait to see how their request is executed.

This relationship of witnessing digitally generated media through a confusing mixture of work and play with a computer interface is exemplified by the satirical work of comedy duo Tim (Heidecker) and Eric (Wareheim), during season 5 of their Adult Swim TV series, Tim and Eric Awesome Show: Great Job! (2006-2010). Episode 10 features a sketch starring actor Paul Rudd, called Celery Man, made viral on the internet with memes of Paul Rudd’s dancing.


The scene begins with Paul Rudd as “Paul,” walking loudly across a bridge in a domed virtual environment. When Paul activates a desktop computer, the machine audibly greets him with an AI named “Paul’s Computer” which automatically runs a storied program called the Cinco Identity Generator 2.5. Paul then speaks commands to load the “Celery Man” sequence.  Video content of Paul dancing in different outfits, styles, and audio appears on across multiple media player windows, producing an experimental multimedia performance. The purpose of Paul actions and the content is never fully disclosed.

Operating on a seemingly primitive interface with a futuristic graphical processing, Celery Man is particularly attention grabbing due to today’s increased leveraging of generative AI for making photos and videos, and the embedding of AI functions into computer operating systems. If the interaction we have with computers enables work to be more leisurely due machine labor value harder that is harder to know, is Paul’s activity driven by amusement or is his curiosity an absurd form of work?

Unpacking the mysteriousness of Paul’s work situation, and his Celery Man clips, we can notice a few things. First, we don’t know exactly where Paul is located but it has the trappings of workplace, not a home entertainment experience. His business casual attire, a cup of coffee, and computer station signify working in an office. The echo his footsteps is an industrial metallic sound. The chair and table are steel and black leather are corporate. The workspace is devoid of distracting décor, possessions, or supplies. This environment has all the trappings of a disembodied Virtual Reality space that could symbolize Paul’s all-consuming focus on his technology, or a literal high-tech workspace that is unattainable for the average person. It is not Paul’s work surroundings that represent his work, it is what’s going on the screen of his computer that he treats as work but looks frivolous and unproductive.


Paul’s workday begins by hitting the space bar of the keyboard once, waking up a PC that indicates it belongs to him without a password or other security credential. At certain points Paul just sits and watches his screen, sipping his coffee, waiting to see the results. Although he is prompting the computer to make changes, he is also a spectator anticipating for something exciting to unfold.

The origin of the media shown is also an unknown element. Are the video clips played from saved files? Is it Paul’s likeness pasted onto an animated digital model? In reality, we know that Paul Rudd was videographed doing dances. Therefore, as a fiction, we cannot assume that that Paul is viewing videos of himself that he recorded on camera. We can assume that the Identity Generator created dancing models that look like Paul sourced from some kind of data. The low-grade technology aesthetic of the sketch positions video of Paul Rudd as hyperreal video generated, touching upon the tension between filmmaking and AI as generative video quality is aimed to replicate the quality of a cinema camera.


Considering the dancing Paul’s as being generative AI video content is especially apparent when the computer offers Paul a new sequence it has been working on. The appearance of the new flamboyant dancer named “Tayne” sounds like Paul but is wholly another character. Paul states his satisfaction, “Now Tayne, I can get into,” and sits back and continues watching. It is worth mentioning that the character Tayne himself is a broken word pun for Entertainment. At no point does Paul become completely passive in his work process. By the end of the scene, the probability of Paul’s activity being slacking off increases by his request for NSFW generation of Nude Tayne, and he ignores an incoming call notification from his wife about an emergency, to focus on the all-consuming work.

Paul certainly enjoys seeing himself as his own self-image made over in a novel way. The alternative looks, running variable speeds and resolution, and models occurring simultaneously through multiple media displayed present possibilities for Paul’s manners and appearance. This emergence of the real self-linked to an artificial self-image clearly represents how individuals prefer to use AI to see and hear themselves reflected back.


Another question is if Paul is doing work or merely escaping work? Work in the arts and entertainment, often requires some degree of play to be successful. The inclusion of digital workflows, makes creativity faster. Tim and Eric’s production team certainly used digital tools to produce experimental media by playing with audiovisual qualities through distortion, editing, and animation. Celery Man functions almost like a love letter for the free expression of digital film and video production. We cannot help but be playful and silly with the output of computers due to their timing and multimedia capabilities.



As to what Paul does for work, the answer is benign. What Paul is working on is reproducing what he already likes, himself, as an entertainer, but for whose benefit? A criticism of AI is that it is most often used not to create new things but to recreate what we already like or believe to be true. That the biases, opinions, ideas, and creations that we already know are just reproduced by being recycled. As consumable media, AI does not present a viewer with new challenges but variations on what is already familiar and well-liked by them and others.

In the case of Paul, the character already likes seeing himself. His narcissism is fed with endless possibilities of how to depict is bodily image, and in the process, he embodies the image processing through interacting with a voice acting and recognizing personal computer. The computer can compile Paul because it functions an extension of him as a tool and service.

Paul is building automated sequences that seem frivolous, while re-familiarizing himself with previous content, and intuitively modifies and recreates processed results in real-time. Paul watches the screen like a show never-before-seen. As computers are automated to perform without human input, AI video retains the traditional screen watching of television. A viewer can sit back and wait for the next visual event to occur. The future of digital labor is a stream of global media objects, prompted and synthesized AI as a companion to play with. The AI company Runway also envisions this kind of relationship with AI video with a streaming service that continuously produces generative visuals.



Human to computer communications is clearly creative through indivisible play and work conditions. Like Cinco Identity Generator 2.5, modern AI models do work by being tooled through users playing with their prompts, with occasional successes and failures. What differs however, is that Celery Man reflects a physical and mental perception of interacting with one’s own electronic image with a digital profile managed by an AI. Paul commands his own entertainment materially with his own data, technology, and thinking.

AI tools are currently in a state of being casual amusement or used for commercial media production. Yet as computer functions are less driven by human mouse and keystrokes, the user becomes a watcher of their own work. Work becomes desire by waiting for a desirable result. which makes work into trouble shooting self-driven amusement in a process that has some level of unpredictability.

Generally, users want AI models to be more stable, predictable, cohesive, and reproducible in a controlled way. Paul certainly gets a stable image of himself, but the medium of the computer interface and video content are affected by the automaticity, shown by an intermixed animation of GUI elements and photographic content.  


While public interest in AI is focused on its product and the medium itself, it is clear that differences in work and play are liberated by the multi-tasking of computers and its multimedia display. So, is Paul now more fascinated by the possibilities of generated content or what the machines capabilities are? Perhaps when AI takes total control of a computer, we can then see the work it is doing. Then we can know just how much really are or are not playing with our computer.