Comedian Garrett Millerick has created an AI avatar modelled on his own material, an experiment that asks whether artificial intelligence can reproduce the timing, persona and improvisational spark that define stand-up comedy. The project has reignited discussion about what audiences value in live performance and how technology might change the craft.
What comedy depends on
Comedy relies on more than words on a page: timing, delivery, audience response and the ability to react in the moment are central. Critics of machine-generated humour argue that many of these elements are inherently human — tied to lived experience, empathy and spontaneous interaction — and may be hard for an avatar to replicate convincingly. Proponents say AI can handle scripted routines and mimic vocal and facial characteristics well enough to entertain, or to expand a performer's reach.
Artistic and practical questions
Millerick's project underscores practical questions about how comedians might use AI tools. Some performers see potential for archiving sets, experimenting with new material, creating virtual appearances or supplementing tours, while others worry about dilution of originality and the loss of work for touring acts. Observers also note that an avatar built from a comedian's own material raises fewer ethical concerns than ones trained on other performers without consent, but broader issues remain.
Legal and ethical considerations
The experiment feeds into ongoing debates over intellectual property, voice and likeness rights, and how creators should be compensated when their work is used to train or power AI systems. Regulators, performers and platforms are still grappling with frameworks to protect artists' rights while allowing technological innovation.
The bigger picture
Millerick's avatar is one of a growing number of experiments at the intersection of entertainment and AI. Whether audiences find a machine-made set genuinely funny, or merely novel, will help shape how the industry adapts. For now, the experiment is less a final verdict than a prompt: it forces comedians, audiences and policymakers to decide what aspects of performance they value enough to preserve, and which they are willing to automate.