GoKawiilis The Verge’s senior AI reporter. An AI beat reporter for more than five years, her work has also appeared in CNBC, MIT Technology Review, Wired UK, and other outlets.
If you’ve got strong creative instincts, the ability to authentically portray emotion, and are capable of staying true to a character’s voice throughout a scene, there’s a job listing calling for your experience.
The catch: You won’t be performing in a theater, a film studio, or an underground performance space. You’d be using your talents to train an AI model for “one of the leading AI companies,” according to the open role posted by Handshake, a company that provides training data to OpenAI and other labs.
Handshake AI is one of a handful of companies of its kind, scrambling to provide more and more niche or specific training data to AI labs in order to feed the models. AI models are often described as “jagged,” meaning they’re typically great at some surprisingly complex tasks but fail deeply at some simple ones. AI companies are trying to fix the gaps in their models’ knowledge with specialized data labeling, and companies like Handshake, Mercor, and Scale AI have adjusted accordingly, hiring professionals in a wide range of industries.
Handshake’s demand for training data tripled last summer, as The Verge reported in December, and the company surpassed a $150 million run rate in November, scrambling to keep up with demand. Handshake and its competitors have touted their networks of tens of thousands (or more) professionals in white-collar industries, from chemists and doctors to lawyers and screenwriters. Many of these professionals worry they’re training AI models in a way that will make their careers obsolete even quicker than may have happened otherwise.
And now the leading AI labs have come for sketch comics, improv actors, and more.
“Handshake AI is inviting actors, improvisers, and performers to join a paid, collaborative improv project to work with one of the leading AI companies,” the job description says, promising participants will be “matched with other performers over video and given a light prompt or scenario to explore together.”
The job listing calls for people with a background in acting, improv, sketch, or theater work of any kind, and it takes pains to imply — multiple times — that it’s looking for people who can essentially “test the limits of the world’s top LLMs’ understanding” by teaching the models how to recognize or replicate human tone and emotions. “Emotional awareness” is one of the requirements, for instance, specifically the “ability to recognize, express, and shift between emotions in a way that feels authentic and human.” The job listing also called for “interactions that feel grounded, human, and fun to play.”
Handshake declined to comment, and the listing does not specify what the training data will be used for.
In recent years, AI companies have leaned fully into “multimodal” models that can not only generate images and video, but also speak with users via voice interactions, complete with realistic inflections. After OpenAI first tested out ChatGPT’s voice mode, the company leaned further into the feature in 2024, when it debuted an Advanced version with a suite of different voices to choose from. Elon Musk’s xAI offers voice chat within Grok. And Anthropic’s Claude has offered a voice feature, at least in beta, since last May.
... continue reading