AI Companies Have Exhausted Human-Made Data. They're Now Using AI to Make More

AI faces three major hurdles right now. The third is the end of available text data on the internet—companies have scraped it all and have yet to find a reliable new source

In part one of this series, I wrote about the global GPU shortage, how it is forcing AI companies to delay plans and change schedules in a lull that may last until 2024, and why they can only sit back and wait for Nvidia to generate more supplies.

In part two, I wrote about the algorithmic stagnation the field has been living in since the great breakthrough of the transformer in 2017, how ChatGPT’s success has accelerated this process, and how companies are trying to improve existing algorithms and create new ones.

In this part, the last of the series, I will explore a third hurdle the industry is facing: the exhaustion of human-made text data from the internet, why ChatGPT-like products aggravate it, and what potential solutions companies are already exploring.