As artificial intelligence models run out of data to train themselves, AI companies are increasingly relying on real humans to create training content.
For years, companies have used gig workers to train AI models on simple tasks like photo identification, data annotation, and labeling. However, as technology advances rapidly, its training requires a more sophisticated workforce.
Companies like Scale AI and Surge AI employ part-time workers with graduate degrees to write essays and creative prompts for the bots to swallow, the New York Times reported. For example, in the past year, Scale AI has been recruiting people who have completed a master's or doctoral degree, are fluent in English, Hindi, or Japanese, and have professional writing experience in areas such as poetry, journalism, or publishing. Job openings have been posted.
What is their mission? To help AI bots “become better writers,” Scale AI said in a post.
And it takes an army of workers to do this kind of work. According to the Times, Scale AI has tens of thousands of contractors working on its platform at any one time.
“What makes AI really useful to users is the human data layer, and it needs to be done by really smart humans, skilled humans, and humans with specific expertise and creative talent. ” said Vice President Willow Primack. He spoke to the New York Times about data operations at Scale AI. “As a result, we have focused specifically on contractors within North America.”
As tech giants scramble to find new data to train their technology, there's a shift toward more sophisticated gig trainers. That's because the program learns incredibly fast and the resources available for learning are already scarce. From scientific papers to news articles to Wikipedia pages, the vast trove of online information is running dry.
Epoch, an AI research institute, has warned that AI data could run out by 2026.
That's why companies are increasingly finding creative ways to ensure their systems never stop learning. Google is considering access to customer data in Google Docs, Sheets, and Slides, while Meta is even considering acquiring publisher Simon & Schuster to assemble its own book collection, Business Insider reports. Previously reported.