AI Startups Race To Train Models On The Only Data Source Left: Fortune Cookies

AUSTIN, TX — With new standards and technological hurdles preventing AI startups from pulling data from the internet, the entire AI industry is now racing to train cutting-edge AI models on the last data source available: fortune cookies.

In order to steer clear of messy litigation, industry leaders scrambled to ink contracts with top fortune cookie writers Lin Jianhao and P. F. Chang.

"It got wild! One agent was even offering up to 7 dollars per fortune!" Jianhao recounted with tears of joy in his eyes.

AI experts say models will have a fraction of the data to pull from, and it remains to be seen what kind of results could come from such training. "We asked ‘FortuneBot 3.0' to write a poem about autumn," said president of AI startup Databricks David Conte as he opened an early beta of the app. "It wrote: ‘Leaves are orange / Like the sun / Blankets and sweaters are comfy." The app then added, "You will meet a long-lost friend."

At publishing time, AI developers had scrambled to solve the issue of racist responses by implementing another AI model trained on the data set of fortune cookies called "WuhanBot" to oversee moderation.