The Data Labeling Company Serving Chinese Model Builders

It’s no secret that the success of today’s large language models wouldn’t be possible without the armies of humans creating the data, like solved math or coding problems, for the AI to be trained on. 

But as important as data labeling companies—such as Scale AI, Turing and Invisible—have been to the AI sector, they’re not absolutely crucial. Their customers have pulled back on their spending from time to time, as my colleague Cory highlighted in this piece from Friday on Scale missing its revenue and profit targets last year. To counteract that, these companies are increasingly turning to new sources of revenue, like new geographies or consulting services for large enterprises. 

For one of the oldest companies in the data labeling industry, Australia-based Appen, that means increasingly serving Chinese model builders.